J. Phys. Chem. A 2006, 110, 2979-2995
2979
Analysis of Photobleaching in Single-Molecule Multicolor Excitation and Fo1 rster Resonance
Energy Transfer Measurements†
Christian Eggeling,*,‡ Jerker Widengren,§ Leif Brand,| Jo1 rg Schaffer,⊥ Suren Felekyan,# and
Claus A. M. Seidel*,#
Department of NanoBiophotonics, Max-Planck-Institute for Biophysical Chemistry, Am Fassberg 11, 37077
Göttingen, Germany, Royal Institute of Technology, AlbaNoVa UniVersity Center, Experimental Biomolecular
Physics, 10691 Stockholm, Sweden, Future Technologies DiVision, VDI Technologiezentrum GmbH,
Graf-Recke-Str. 84, 40239 Düsseldorf, Germany, Carl Zeiss AG, DeVelopment Light Microscopy,
Königsallee 9-21, 37081 Göttingen, Germany, and Lehrstuhl für Molekulare Physikalische Chemie,
Heinrich-Heine-UniVersity Düsseldorf, UniVersitätsstrasse 1, 40225 Düsseldorf, Germany
ReceiVed: August 15, 2005; In Final Form: January 12, 2006
Dye photobleaching is a major constraint of fluorescence readout within a range of applications. In this study,
we investigated the influence of photobleaching in fluorescence experiments applying multicolor laser as
well as Förster resonance energy transfer (FRET) mediated excitation using several red-emitting dyes frequently
used in multicolor experiments or as FRET acceptors. The chosen dyes (cyanine 5 (Cy5), MR121, Alexa660,
Alexa680, Atto647N, Atto655) have chemically distinct chromophore systems and can be excited at 650 nm.
Several fluorescence analysis techniques have been applied to detect photobleaching and to disclose the
underlying photophysics, all of which are based on single-molecule detection: (1) fluorescence correlation
spectroscopy (FCS) of bulk solutions, (2) fluorescence cross-correlation of single-molecule trajectories, and
(3) multiparameter fluorescence detection (MFD) of single-molecule events. The maximum achievable
fluorescence signals as well as the survival times of the red dyes were markedly reduced under additional
laser irradiation in the range of 500 nm. Particularly at excitation levels at or close to saturation, the 500 nm
irradiation effectively induced transitions to higher excited electronic states on already excited dye molecules,
leading to a pronounced bleaching reactivity. A theoretical model for the observed laser irradiance dependence
of the fluorescence brightness of a Cy5 FRET acceptor dye has been developed introducing the full description
of the underlying photophysics. The model takes into account acceptor as well as donor photobleaching from
higher excited electronic states, population of triplet states, and energy transfer to both the ground and excited
states of the acceptor dye. Also, photoinduced reverse intersystem crossing via higher excited triplet states is
included, which was found to be very efficient for Cy5 attached to DNA. Comparing continuous wave (cw)
and pulsed donor excitation, a strong enhancement of acceptor photobleaching by a factor of 5 was observed
for the latter. Thus, in the case of fluorescence experiments utilizing multicolor pulsed laser excitation, the
application of the appropriate timing of synchronized green and red laser pulses in an alternating excitation
mode can circumvent excessive photobleaching. Moreover, important new single-molecule analysis diagnosis
tools are presented: (1) For the case of excessive acceptor photobleaching, cross-correlation analysis of singlemolecule trajectories of the fluorescence signal detected in the donor and acceptor detection channels and
vice versa shows an anticorrelated exponential decay and growth, respectively. (2) The time difference, Tg Tr, of the mean observation times of all photons detected for the donor and acceptor detection channels within
a single-molecule fluorescence burst allows one to identify and exclude molecules with an event of acceptor
photobleaching. The presented single-molecule analysis methods can be constrained to, for example, FRETactive subpopulations, reducing bias from FRET-inactive molecules. The observations made are of strong
relevance for and demand a careful choice of laser action in multicolor and FRET experiments, in particular
when performed at or close to saturation.
Introduction
Recent years have witnessed a strongly increased number of
biological, physical, and chemical applications based on fluo†
Part of the special issue “Jürgen Troe Festschrift”.
* To whom correspondence should be addressed. E-mail: [email protected]
(C.E.); [email protected] (C.A.M.S.).
‡ Max-Planck-Institute for Biophysical Chemistry.
§ AlbaNova University Center.
| VDI Technologiezentrum GmbH.
⊥ Carl Zeiss AG.
# Heinrich-Heine-University Düsseldorf.
rescence detection. Due to the superior sensitivity to environmental properties as well as its multidimensionality, that is, its
ability to provide various simultaneous readouts (e.g., intensity,
anisotropy, lifetime, and spectral characteristics), fluorescencebased detection has developed into one of the most important
tool categories in life science.1 Recent technological developments have enabled an ultimate sensitivity in fluorescence
detection, including the possibility to observe and analyze single
fluorescent molecules.2-12 Thereby, molecular properties otherwise hidden by ensemble averaging can be revealed. The
10.1021/jp054581w CCC: $33.50 © 2006 American Chemical Society
Published on Web 02/16/2006
2980 J. Phys. Chem. A, Vol. 110, No. 9, 2006
benefits of a single-molecule perspective have been taken
advantage of in many applications, ranging from observations
in solids and solutions to live-cell imaging. To go beyond mere
localization of a molecule in a three-dimensional space, the
single molecules under study are often characterized not only
with respect to intrinsic molecular parameters such as the
fluorescence lifetime and anisotropy but also with respect to
information from the fluctuations of the determined parameters.13 This information can often contain substantial additional
and complementary information about, for example, molecular
properties, mobilities, interactions, and dynamics. Fluorescence
correlation spectroscopy (FCS),14 burst-size distribution analysis,15 photon count histogram (PCH) analysis,16 and fluorescence
intensity distribution analysis (FIDA)17 belong among the most
prominent fluorescence fluctuation spectroscopies. FCS is based
on the analysis of fluorescence fluctuations in time, while PCH
and FIDA are based on fluctuating amplitude analysis. Further
development of FCS, FIDA, and PCH resulted in their multicolor successors fluorescence cross-correlation spectroscopy
(FCCS)18 and two-dimensional FIDA or PCH.19,20
An ultimate development in single-molecule spectroscopy is
represented by single-molecule multiparameter fluorescence
detection (MFD), which in addition to the spatial information
permits the simultaneous collection of all traditional fluorescence
parameters, that is, intensity, lifetime, and anisotropy in two or
more spectral ranges.21,22 Multiparameter single-molecule analysis makes it possible to establish two- or even multidimensional
fluorescence parameter histograms, for example, of fluorescence
brightness, lifetime, anisotropy, or spectral information.23-31
Such multidimensional histograms enable the exposure, sorting,
and specific selective analyses of different subpopulations within
a heterogeneous sample22,24,28-32 and have been successfully
applied to study the structure and function of biomolecules in
high detail.29,30,33 Similarly, for more robust and high-throughput
applications, direct fitting of such histograms has been accomplished by 2D-FIDA,19 two-color PCH,20 or photon arrivaltime interval distribution (PAID).34
The latter techniques as well as MFD provide good examples
of the virtues in single-molecule and fluctuation studies of
applying two or more fluorescence labels emitting in different
spectral ranges. For example, monitoring the coincidence of
multiple fluorescence colors in space or time can unravel
binding, transport, and localization events with high specificity
and in large detail.18,19 As a multicolor method, Förster
resonance energy transfer (FRET)35 provides a sensitive tool
for measuring inter- and intramolecular distances and distance
changes in a range of 1-10 nm. FRET makes use of two labels,
a donor and an acceptor, where the excitation of the donor dye
can result in acceptor fluorescence emission due to a transfer
of the donor’s energy to the acceptor. The efficiency of the
energy transfer and thus the amount of fluorescence emitted by
the acceptor dye strongly depends on the distance between the
two dyes. The transfer efficiency also depends on characteristics
such as the spectral overlap and the orientation of the two labels.
A large number of life science applications are based on
FRET,27,36-41 and FRET has also been combined with fluorescence fluctuation spectroscopy42-44 and single-molecule spectroscopy.29-32,39,45 In the latter case, methods such as MFD disclose
different molecular states based on their specific FRET efficiencies.29,30 Additionally, application of alternating-laser excitation
(ALEX) makes it possible to set up two-dimensional histograms
displaying FRET efficiencies and donor and acceptor labeling
degrees, distinguishing also low-FRET events from donor-only
labeled species.31 Pulsed interleaved excitation (PIE)46 can reveal
Eggeling et al.
species with low FRET efficiencies in a similar fashion. Here,
spectral cross-talk is eliminated via alternating excitation pulses
of different colors, one for excitation of the donor dye and for
FRET, and the other for separate excitation of the acceptor dye,
respectively.
High signal levels are often important in fluorescence
experiments, to acquire the necessary information content, to
reach the required image scanning rates, and/or to reach the
needed time resolution. This is especially pronounced for singlemolecule or fluorescence fluctuation experiments, where a low
signal cannot be compensated by a concentration increase of
the fluorescent molecules investigated. As a consequence, high
laser irradiances are often applied. However, the maximum
applicable irradiance is limited by the photophysics of the dye
labels, in particular by singlet and triplet saturation as well as
by photobleaching. Photobleaching is to a large extent a
consequence of an enhanced reactivity of the fluorophore in its
excited state,47 and it thus scales with the applied irradiance.
Moreover, recent work on several dyes has shown that even
for continuous wave (cw) excitation and in particular for shortpulsed laser excitation, photobleaching is severely enhanced due
to the population of higher excited electronic states.48,49 In polar
solvents such as water, these states couple quite efficiently with
ionic states, leading to a pronounced reactivity.50,51 This
nonlinear dependence on the applied irradiance requires wellchosen experimental conditions with respect to laser irradiance
and pulse length. Eventually, optimal conditions are a tradeoff
between maximum observation time, maximum photon number,
and minimized (heterogeneous) photobleaching. Nonetheless,
this optimization is crucial and is strongly facilitated by detailed
knowledge of the relevant photophysical mechanisms.
In this work, we focus on characterizing the photobleaching
in two-color and FRET experiments, where laser light either of
several wavelengths and/or of a wavelength not usually employed can generate additional photophysical transitions within
the dyes. For example, in most FRET experiments, laser light
of only one wavelength is required for both, direct excitation
of the donor and FRET-mediated excitation of the acceptor dye.
However, effects from direct irradiation at this wavelength often
have to be taken into consideration also for the acceptor dye,
although the laser light does not excite the acceptor directly, at
least not from its ground state. This also holds true for multicolor
experiments, where several different dye labels and laser
wavelengths are employed simultaneously. Some FRET imaging
techniques, such as photobleaching FRET (pbFRET), even
directly exploit the specific photodestruction of donor or
acceptor dyes to gain more spatial information on molecular
distributions and mobilities in cells.52-54 However, as in most
fluorescence experiments, photobleaching of either dye is
unwanted, since it usually leads to limited and biased
results.43,47-49,55,56 Recently, photodestruction-intermediate states
of a FRET acceptor have been reported, which are nonemitting
but still able to quench the fluorescence of the donor.57 On the
other hand, the sensitivity of the red-excited and red-emitting
dye cyanine 5 (Cy5) to green light has also been reported. In
these experiments in deoxygenated aqueous solutions and with
added triplet quencher, excitation at 633 nm drove the dye into
a nonfluorescent, quasi-photobleached state, while addition of
green light at around 500 nm led to recovery of the Cy5
fluorescence, turning Cy5 into an optical switch.58,59 In view
of these seemingly contradictive results, the overall importance
of dye photobleaching, and the additional effects observed under
multiwavelength and FRET-mediated excitation, a more detailed
and general understanding of the photobleaching impact of green
Influence of Photobleaching in FRET Experiments
light for red-emitting dyes is needed, as is the development of
methods providing remedies to these photobleaching effects.
In this study, the photobleaching properties of several redemitting dyes with chemically distinct chromophore systems,
Cy5, MR121, Alexa660, Alexa680, Atto647N, and Atto655, and
of a Cy5 FRET conjugate were investigated in solution at
simultaneous or separate green and red laser excitation using
FCS and single-molecule analysis methods. The dyes belong
to various classes of, for example, oxazine, carbopyronine, or
cyanine chromophores. Especially Alexa660 and 680 show very
broad absorption spectra and are expected to be distinct from
the other dyes. It is outlined how extensive photobleaching by
the green laser light, interacting with the excited states of the
dyes, limits the maximum accessible fluorescence brightness.
A clear distinction between the cw and pulsed excitation modes
can be observed, and the features of photobleaching with respect
to the population of the triplet state and of higher excited
electronic states, energy transfer to the acceptor’s ground and
excited states, as well as photoinduced reverse intersystem
crossing via higher excited triplet states are considered. The
presented results are to a great extent only accessible by singlemolecule fluorescence analysis. They reveal major limitations
in multicolor and FRET experiments that can be attributed to
enhanced photobleaching of the acceptor dye. On the basis of
the findings, optimal experimental conditions with respect to
fluorescence output and minimized photobleaching in terms of
laser irradiance, excitation mode, and pulse synchronization of
different laser wavelengths are outlined. Moreover, readily
accessible new single-molecule analysis methods are presented
to check for the amount of photobleaching, providing adequate
guidance for the proper optimization adjustments: (1) For the
case of excessive acceptor photobleaching, cross-correlation
analysis of single-molecule trajectories of the fluorescence signal
detected in the donor and acceptor detection channels and vice
versa shows an anticorrelated exponential decay and growth,
respectively. (2) The time difference, Tg - Tr, of the mean
observation times of all photons detected for the donor and
acceptor detection channels within a single-molecule fluorescence burst allows one to identify and exclude molecules with
an event of acceptor photobleaching. In this way, various singlemolecule analysis methods presented above can be constrained
to, for example, FRET-active subpopulations, reducing bias from
FRET-inactive molecules.
Theory
The theoretical part introduces the photokinetic model of
photobleaching in two-color and FRET experiments. The first
part gives a general theoretical characterization of the fluorescence emission of a dye regarding photobleaching. In the second
part, this theoretical description is extended to Förster resonance
energy transfer (FRET) experiments.
General Photokinetic Model of Photobleaching for Direct
One- or Two-Color Excitation. Figure 1A depicts the general
photophysical model of photobleaching of a dye, based on five
electronic levels: the singlet ground state, S0, the first excited
singlet state, S1, the lowest triplet state, T1, and higher excited
singlet and triplet states, Sn and Tn. Vibrational substates can
be disregarded due to their comparatively short lifetime. The
kinetic rate constants underlying the possible transitions between
the electronic states are denoted as follows: k01, excitation from
S0 to S1; k10, sum of all processes leading to the de-excitation
of S1 to S0 (i.e., internal conversion, kIC, fluorescence, kf); kISC,
intersystem crossing (ISC) from S1 to T1; kT, rate of back ISC
from T1 to S0 (inverse triplet lifetime); kS1n and kT1n, excitation
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2981
Figure 1. Photokinetic model of photobleaching. (A) Energy diagram
of a dye molecule including five electronic levels and their transition
rate constants. For details, see text. (B) Simulated single-molecule time
trajectory of the fluorescence brightness, F1(t), in the donor channel
(green line) and in the acceptor channel (red line) over time, t,
exemplified for the event of acceptor photobleaching. DF1 (donor
brightness in the presence of FRET) and AF1 (acceptor brightness) depict
the respective values of brightness before the photobleaching event,
and D0F1 (donor brightness in the absence of FRET) and R D0F1 (crosstalk of donor fluorescence into the acceptor detection channel) depict
the respective values of brightness after the photobleaching event.
“Start” and “end” denote the beginning and the end of the singlemolecule transit through the detection volume, and “bleach” denotes
the incident of the acceptor photobleaching. The time tz represents the
average time before photobleaching, given by tz ) 1/Akz (eq 6), and
the time tt represents the average transit time of a single molecule
through the detection volume. The parameters Tg and Tr exemplify the
medial time of fluorescence emission of both detection channels, as
used for the Tg - Tr analysis in Figure 6.
from S1 to Sn or T1 to Tn, respectively; kSn1 and kTn1, de-excitation
/
/
and kReISC
,
of Sn and Tn to S1 and T1, respectively; and kISC
crossing between Tn and Sn, that is, photoinduced ISC and
reverse ISC. Photobleaching reactions occur from S1, T1, Sn,
and Tn and are accounted for by the rate constants kbS1, kbT1,
/
/
, and kbTn
, respectively. The sum of all depopulation
kbSn
processes of S1, k0 ) kISC + kIC + kf, gives the inverse
fluorescence lifetime, τ-1.
The rate constants for excitation to S1, Sn, and Tn involve the
absorption of a photon and thus depend on the applied laser
irradiance, I.
kX(t) ) σX(λ) γI(t)
(X ) 01, S1n, T1n)
(1)
σ01(λ), σS1n(λ), and σT1n(λ) are the corresponding absorption
2982 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Eggeling et al.
cross sections, λ is the excitation wavelength, and γ ) λ/(hcl)
is the reciprocal photon energy, with h being Planck’s constant
and cl being the velocity of light. I and σX are given in units of
watts per square centimeter and square centimeters, respectively.
The time, t, dependence of kX(t) results from I(t), which is
constant for continuous wave (cw) excitation and follows the
pulse shape and the repetition rate in the case of pulsed
excitation.
Those processes that involve the excitation to the higher excited electronic states, Sn and Tn, that is, photoinduced ISC and
reverse ISC and photobleaching, can be expressed by composite
rate constants starting from S1 and T1, introducing effective cross
sections, σISC-n, σReISC-n, and σbXn and, given in cm2 W-1 s-1.
pulses for the present experimental conditions of ∆t ≈ 0.28 ms
and υ ) 73.5 MHz. As shown previously,49 for simplification,
a rectangular pulse shape with the experimentally determined
width of in this case 180 ps can be assumed.
For cw excitation, a more simple approach can be used,
applying the steady state populations of the first excited singlet
state, S1eq, and of the lowest triplet state, T1eq, deduced from eq
3.
S1eq )
(σS1nγI(t))φISC-n ) σISC-nI(t)
photoinduced reverse ISC: kReISC-n(t) )
(σT1nγI(t))φReISC-n ) σReISC-nI(t)
The parameters σX1n are the previously defined absorption cross
sections for excitation from S1 to Sn (X ) S) and T1 to Tn (X )
/
/
/kSn1, φReISC-n ) kReISC
/kTn1, and φbXn
T) (eq 1), and φISC-n ) kISC
/
) kbXn/kXn1 are the quantum yields for ISC, reverse ISC, and
photobleaching from Sn and Tn.
By introducing these effective cross sections (eq 2), the
populations of the higher excited states do not need to be
explicitly calculated. This is allowed, since these states have a
comparatively low lifetime of kSn1-1, kTn1-1 < picoseconds.48
This leaves us with a rate equation system involving only three
electronic states: the singlet ground, first excited singlet, and
lowest triplet states.
Ṡ0(t) ) -k01(t)S0 + k10S1 + kTT1
(3)
Ṡ1(t) ) k01(t)S0 [k10(t) + kISC + kISC-n(t) + kbS1 + kbSn(t)]S1 + kReISC-n(t)T1
Ṫ1(t) ) [kISC + kISC-n(t)]S1 [kT + kReISC-n(t) + kbT1 + kbTn(t)]T1
The time-dependent populations of each electronic level can
be calculated by solving the rate equation system for a given
time course of the irradiance, I(t) (compare eqs 1 and 2), and
with initial populations of S0(0) ) 1 and S1(0) ) T1(0) ) 0.
The calculation of S1(t) on the other hand allows for the
quantification of the fluorescence brightness, F1(t) ) ΨΦτ-1S1(t),
which expresses the average count rate detected for a single
dye molecule. Ψ denotes the fluorescence detection efficiency
of the experimental setup and Φ the fluorescence quantum yield
of the dye.
In the case of pulsed excitation with repetition rate υ, the
brightness is given by the populations in the S1 state after the
ith excitation pulse, S1(i).49
nP
(with nP ) ∆tυ)
(4)
Here, ∆t denotes the observation time window and nP ) ∆tυ
the number of laser pulses within ∆t. nP amounts to about 20 580
(5)
Photobleaching can be constituted into this steady state approach
by an effective bleaching rate constant, kz, and the probability,
pintact(t), of still being fluorescent after time t.48,49
kz ) (kbS1 + kbSn)S1eq + (kbT1 + kbTn)T1eq )
photobleaching, X ) S or T: kbXn(t) )
(σX1nγI(t))φbXn ) σbXnI(t) (2)
∑i S1(i)/∆t
k01(kT + kISC + kISC-n + kReISC-n) +
k10(kT + kReISC-n) + kT(kISC + kISC-n)
T1eq ) (kISC + kISC-n)/(kT + kReISC-n)S1eq
photoinduced ISC: kISC-n(t) )
F1 ) ΨΦ
k01(kT + kReISC-n)
eff
[keff
b1 + σbn I(t)]S1eq
pintact(t) ) exp(-kzt)
(6)
with
keff
b1 ) [1 + (kISC + kISC-n)/(kT + kReISC-n)]kbX1
σeff
bn ) [1 + (kISC + kISC-n)/(kT + kReISC-n)]σbXn
kbX1 is the rate constant of photobleaching from S1 and T1, while
kbXn ) σbXnI(t) depends on the irradiance, I(t), and represents
the composite rate constant of higher-order photobleaching via
Sn and Tn (eq 2). In our experiments, the rate constants, kbX1
and kbXn, cannot be assigned to specifically belong to either the
singlet system (X ) S) or the triplet system (X ) T) and are
therefore, for simplification, assumed to have the same magnitude for X ) S or T. This simplification introduces the effective
rate constant of first-order photobleaching, keff
b1 , and the effective cross section of higher-order photobleaching, σeff
bn , after the
incident of excitation to S1.
The average brightness, F1, within the observation time
window, ∆t, is in the case of cw excitation thus given by eq 7.
1
F1 ) ΨΦτ-1S1eq
[1 - exp(-kz∆t)]
kz∆t
(7)
For a single molecule diffusing through the focal observation
volume, the observation time window is given by the mean
transit time, ∆t ) tt.
Photobleaching in Fo1 rster Resonance Energy Transfer
(FRET) Experiments. FRET defines the dipole-dipole, nonradiative energy transfer from an excited dye molecule, D*
(donor), to another dye molecule, A (acceptor). The energy
transfer leads to the electronic excitation of the acceptor, A*,
and the de-excitation of the donor, D* + A f D + A*.35 Thus,
a bichromophoric system of donor and acceptor dye has to be
considered. To distinguish between donor and acceptor, each
dye is denoted by the index “D” or “A” throughout. In our case,
we distinguish between three different energy transfer mechanisms: energy transfer from the donor’s first excited singlet
state, DS1, to the acceptor’s (i) singlet ground state, AS0, (ii)
first excited singlet state, AS1, and (iii) first excited triplet state,
AT .
1
Influence of Photobleaching in FRET Experiments
ES0
(i) DS1 + AS0 98 DS0 + AS1
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2983
(8)
ES1
(ii) DS1 + AS1 98 DS0 + ASn
ET1
(iii) DS1 + AT1 98 DS0 + ATn
The three pathways correspond to competitive FRET mechanisms previously observed in single bichromophoric molecules.60,61 Energy transfer pathways starting from the lowest
triplet state or from higher excited electronic states of the donor
can be neglected. Due to the lower energy of DT1 and the much
lower lifetimes of DSn and DTn compared to DS1, energy transfer
pathways starting from DT1, DSn, and DTn are much more insufficient. In all three energy transfer pathways of eq 8, donor fluorescence is quenched and, without direct laser excitation, the
corresponding excited states of the acceptor are populated. To
account for this quenching of donor fluorescence, we differentiate between donor in the presence (index “D”) and donor in
the absence (index “D0”) of FRET to an acceptor dye. Pathway
i sensitizes acceptor fluorescence and is, thus, the most common
energy transfer mechanism, usually desired in FRET experiments. The efficiency of each energy transfer mechanism can
be defined by the energy transfer efficiencies, ES0, ES1, or ET1.
EX )
kFRET-X
∑kFRET-X + D0k0
(X ) S0, S1, T1)
(9)
) 1/τD0 is the inverse fluorescence lifetime of the donor
in the absence of any FRET to an acceptor, and ∑ denotes the
sum over all energy transfer pathways. The microscopic rate
constants, kFRET-X, for the energy transfer are given (a) by the
spectral overlap of the donor fluorescence emission with the
acceptor’s ground (X ) S0), the first excited singlet (X ) S1),
or the lowest triplet state (X ) T1) absorption spectrum, (b) by
the relative orientation of the donor and acceptor dyes, and (c)
by the spatial distance, RDA, between both dyes, kFRET-X ∼
RDA-6.35 This characteristic sixth-order dependence on RDA
makes FRET a sensitive tool to measure distance changes,
usually in a range of 1-10 nm.
Some of the previously defined rate constants (eqs 1 and 2)
have to be adapted to the occurrence of FRET.
D0k
0
(a) donor fluorescence lifetime
τD-1(t) ) Dk0(t) )
D0
lifetime of the donor in the absence of an energy transfer to an
acceptor dye, and Aσ01, the absorption cross section for direct
laser light excitation of the acceptor to AS1. Aσ01 is usually very
low, since the wavelength of the excitation light is optimized
for the donor and far away from the absorption maximum of
the acceptor. AσISC-n,AσReISC-n, and AσbXn are the composite cross
sections of the acceptor for photoinduced ISC and reverse ISC
and photobleaching induced by the time-dependent irradiance,
/
/
I(t) (eq 2), while AφISC-n
, AφReISC-n
, and Aφ/bXn express the
quantum yields for photoinduced ISC and reverse ISC and
photobleaching from the higher excited states of the acceptor
induced by FRET (eq 2). The asterisk marks the fact that, due
to the different spectral ranges of the laser- and FRET-induced
excitation, these quantum yields may differ from those defined
for the corresponding processes following direct absorption of
the excitation light. The latter quantum yields, AφISC-n,
Aφ
A
ReISC-n, and φbXn, are included within the composite cross
A
sections σISC-n,AσReISC-n, and AσbXn (eq 2). Due to the Stokes
shift between excitation and emission of the donor, the energy
of the laser light for donor excitation is higher than the energy
mediated by FRET. This results in higher energy levels of ASn
and ATn and, thus, higher efficiencies for ISC, reverse ISC, and
photobleaching from these states, compared to FRET-mediated
excitation.
From eq 10, a decreased fluorescence lifetime of the donor
dye as well as the sensitization of acceptor fluorescence follows
upon occurrence of FRET. Furthermore, eq 10 also takes into
account the fact that a certain energy transfer cannot occur if
any of the involved initial states are not populated. Thus, the
rate of FRET scales with the population of the first excited
singlet state of the donor (DS1(t)) or with those of the
corresponding states of the acceptor (AS0(t), AS1(t), AT1(t)).
Full Simulation of the Donor-Acceptor System. To
simulate the fluorescence signal of the bichromophoric donoracceptor system, the population of the donor’s and acceptor’s
first excited singlet state, DS1(t) and AS1(t), can be calculated
by simultaneously solving the rate equation system of eq 3 for
donor and acceptor dye. The complete rate equation system in
the case of FRET thus includes six equations, three for each
dye, which are linked via eq 10. Fluorescence emitted by the
donor and acceptor is detected in two spectrally distinct detection
channels. The corresponding fluorescence brightnesses, gF1 and
rF , detected in the “green” and “red” detection channels,
1
respectively, are given by the fluorescence brightness of the
donor, DF1, and of the acceptor, AF1.
F 1 ) DF 1
g
k0 + AS0(t)kFRET-S0 + AS1(t)kFRET-S1 + AT1(t)kFRET-T1
r
(b) acceptor excitation rate
k01(t) ) Aσ01γI(t) + DS1(t)kFRET-S0
A
F1 ) ΨgΦDτD-1 DS1(t) and
D
/
kISC-n(t) ) AσISC-nI(t) + DS1(t)kFRET-S1AφISC-n
A
(d) acceptor photoinduced reverse ISC
/
kReISC-n(t) ) AσReISC-nI(t) + DS1(t)kFRET-T1AφReISC-n
A
(e) higher-order acceptor photobleaching (X ) S or T)
/
kbXn(t) ) AσbXnI(t) + DS1(t)kFRET-X1AφbXn
D0k
0
)
D0k
10
F 1 ) AF 1 + R DF 1
with
(c) acceptor photoinduced ISC
A
(11)
(10)
+ DkISC ) τD0-1 depicts the inverse fluorescence
A
F1 ) ΨrΦAτA-1 AS1(t)
Ψg and Ψr are the detection efficiencies of each detection
channel, ΦD and ΦA are the fluorescence quantum yields, and
τD and τA are the fluorescence lifetimes of donor and acceptor,
respectively. The donor cross-talk, R, amounts for the fact that
cross-talk between the two detection channels may occur due
to fluorescence emission of the donor bleeding into the
acceptor’s (red) detection channel. DF1 and AF1 can be calculated
from DS1(t) and AS1(t) corresponding to eq 4 for pulsed excitation
and eq 7 for cw excitation.
Inhomogeneous Focal Volume. For a focal volume formed
by an objective lens of high numerical aperture, the laser
2984 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Eggeling et al.
irradiance and, thus, all irradiance-dependent rate constants
strongly depend on space. The spatial dependence of the focal
irradiance can be approximated by a three-dimensional Gaussian
profile, I(x,y,z) ) I0 exp(-2(x2 + y2)/ω02) exp(-2z2/ωz2), with
the focal irradiance I0 ) I(z ) 0) and 1/e2 radii ω0 and ωz in
radial (x, y) and axial (z) directions, respectively. To simplify
expressions for photophysical processes, it is justified to
approximate the excitation volume by a rectangular profile with
radius ω0. As shown before, an average rate constant, k(I), can
be defined in such a case, where I is given by I0/2.24,48,49,62,63
Fluorescence Correlation Spectroscopy (FCS). The overall
fluorescence signal detected from a solution of fluorescent
molecules is given by the mean number of molecules in the
observation volume, c, denoted concentration throughout, and
by the fluorescence brightness of each single molecule, F1 (see
eqs 4 and 7); F ) c × F1. Variations in c or in F1 are caused
by, for example, diffusion and chemical reaction. These variations lead to temporal fluctuations in the fluorescence signal
around an average value, F(t) ) 〈F(t)〉 + δF(t), which are
analyzed in FCS by means of the second-order autocorrelation
function, G(tc).13
G(tc) ) 1 +
〈δF(t) δF(t + tc)〉
(12)
〈F(t)〉2
Here, tc represents the correlation time. Triangular brackets
indicate averaging over the measurement time, t.
Irreversible photobleaching reactions to a nonfluorescent
product give rise to an abrupt termination of fluorescence
emission, and thus to characteristic fluorescence fluctuations.
In the autocorrelation function, the abrupt termination of
fluorescence emission results in an apparently shorter observation time. Because the total sample volume is much larger than
the excitation volume, the depletion of fluorophores within the
focus due to photobleaching will be balanced by a net in-flow
from out-of-focus regions due to the formed concentration
gradient. Consequently, photobleaching can be treated as a
chemical pseudoequilibrium reaction using the effective bleaching rate constant, kz(I0/2) (eq 6). In this way, the autocorrelation
function taking diffusion dynamics, dark state population (such
as the triplet state), and photobleaching kinetics into account
can be approximated by eq 13.48,49,62,63
1
G(tc) ) 1 + [GDiff(tc) GB(tc) + GDark(tc)]
c
(13)
with
GDiff(tc) )
(
)(
1
1
1 + tc/τDiff 1 + (ω /ω )2(t /τ )
0 z
c Diff
GB(tc) ) 1 - A + A exp(-kztc)
GDark(tc) )
FD
exp(-tc/τDark)
1 - FD
)
1/2
(diffusion)
(bleaching)
(dark state)
The characteristic diffusion time is given via the diffusion
coefficient, D, by τDiff ) ω02/(4D). c is the concentration, that
is, the mean number of fluorescent molecules in the detection
volume, FD denotes the equilibrium fraction of molecules in
the dark state such as T1, τDark is the characteristic dark state
correlation time, characterized by the population and depopulation kinetics, and A is an amplitude describing the average
bleaching decay with rate constant kz. While the dark state term,
GDark(tc), is added, the bleaching, GB(tc), and the diffusion terms,
GDiff(tc), are multiplied, since they both decay in about the same
time scale.
Cross-Correlation Analysis of Acceptor Photobleaching
in Single-Molecule Bursts. In the present FRET experiments,
the distinct detection of donor and acceptor fluorescence in the
two detection channels permits the calculation of the crosscorrelation function of both signals.
Ggr(tc) ) 1 +
Grg(tc) ) 1 +
〈δgF1(t) δrF1(t + tc)〉
〈gF1(t)〉〈rF1(t)〉
〈δrF1(t) δgF1(t + tc)〉
〈gF1(t)〉〈rF1(t)〉
(14)
gF
r
1 and F1 denote the fluorescence brightness detected in the
donor and acceptor detection channels (eq 11). Note that, in
contrast to eq 12, the cross-correlation function is in this case
defined on the basis of the fluctuating fluorescence brightness,
F1(t), since in the present experiments the cross-correlation
analysis of the acceptor photobleaching is solely applied to time
trajectories of single molecules.
The analysis of the cross-correlation function discloses
processes that influence both the donor and acceptor fluorescence signals simultaneously (coincidence signal).18 Three
processes reveal concomitant variations in the fluorescence
brightness of both dyes and thus influence the cross-correlation
function. (1) Fast fluctuations in the microsecond time domain
occur due to the population of dark states such as the triplet
state of both dyes62 or photoisomerization.64-66 The population
of a donor’s dark state interrupts the energy transfer and thus
the fluorescence emission of the acceptor. As discussed before
(eq 8), energy transfer from the donor’s dark (triplet) state is
negligible. Further, the acceptor in the triplet or dark state may
absorb the donor’s energy in a different way as in the ground
state, leading to concomitant change in donor fluorescence
emission. Due to this correlated behavior, these dark state
fluctuations contribute to the cross-correlation function in a
similar way as the dark state term, GDark(tc), of eq 13. (2) Since
both dyes are tagged to the same molecule, diffusion through
the focus with a mean transit time, tt, influences the fluorescence
emission of both dyes in the same way. (3) Irreversible
photobleaching of the acceptor dye makes it unable to emit
fluorescence and to absorb the donor’s energy. Acceptor
photobleaching thus reveals an interruption of FRET accompanied by a loss of acceptor fluorescence and a concomitant
rise in fluorescence brightness of the donor, exemplified in
Figure 1B. This photobleaching step occurs after a time, tz, that
is on average given by the effective bleaching rate constant of
the acceptor; tz ) 1/Akz (eq 6). Photodestruction-intermediate
states that are nonemitting but still able to quench the fluorescence of the donor have been observed in previous singlemolecule FRET experiments57 but have not been observed in
our experiments and are thus neglected.
As given in the Supporting Information, Ggr(tc) and Grg(tc)
can be approximated for the special case of tz , tt, that is, if
photobleaching is faster than diffusion.
Ggr(tc) ) grB0 + grB1 exp(-kztc) + GDark(tc)
Grg(tc) ) rgB0 - rgB1 exp(-kztc) + GDark(tc)
with
(15)
Influence of Photobleaching in FRET Experiments
gr
R D0F1 tt
B0 ≈ A
F tz
gr
R D0F1 tt
A
F tz
rg
D
1
B0 ≈ 1 +
rg
1
F1
B1 ≈ D0
F1
D
F1
B1 ≈ 1 - D0
F1
where, according to Figure 1B, DF1 and AF1 represent the
fluorescence brightness of donor and acceptor before the
photobleaching event, that is, in the presence of FRET, and D0F1
represents the brightness of the donor after the photobleaching
event, with R D0F1 being the concomitant brightness detected
in the acceptor channel due to cross-talk, R. The most important
criterion for photobleaching is the fact that the correlation
functions Ggr(tc) and Grg(tc) have an opposite time dependence;
Ggr(tc) shows an exponential decay and Grg(tc) an exponential
growth, because D0F1 > DF1. The validity of eq 14 is based on
the condition that the contribution of diffusion to the crosscorrelation data (similar to GDiff(tc) in eq 13) can be neglected.
Otherwise, the data analysis would be significantly complicated.
Materials and Methods
Samples. The dyes rhodamine 110 (Rh110, excitation
maximum 500 nm, emission maximum 525 nm; Radiant Dyes,
Wermelskirchen, Germany), MR121 (excitation maximum 650
nm, emission maximum 665 nm; Roche Diagnostics, Mannheim,
Germany), cyanine 5 (Cy5, excitation maximum 650 nm,
emission maximum 670 nm; Molecular Probes, Eugene, OR),
Alexa660 (excitation maximum 660 nm, emission maximum
685 nm; Molecular Probes), Alexa680 (excitation maximum 680
nm, emission maximum 700 nm; Molecular Probes), Atto655DNA (5′-d(CGG CCT ATT AGA TAT TTA TTG CTA
T(Atto655) TA CCA, free dye: excitation maximum 665 nm,
emission maximum 685 nm; Atto-Tec, Siegen, Germany), and
Atto647N-DNA (5′-d(TTG ATT CGG TCT ATG CAA AAA
AA T(Atto647N) TGC ATA GAG GAT TGC AA, free dye:
excitation maximum 645 nm, emission maximum 670 nm; AttoTec) were used for control measurements. Stock solutions of
these dyes were diluted in double-distilled water to a final
concentration of ≈10-9 M for FCS and ≈10-12 M for singlemolecule studies.
A double-stranded DNA molecule labeled with Rh110 or
Alexa488 (excitation maximum 500 nm, emission maximum
524 nm; Molecular Probes) and Cy5 was applied for FRET
measurements, denoted FRET conjugate throughout. The (-)strand 5′-d(GCA TCG ATC CTC ATT AT-A AGT) was labeled
with Alexa488 or Rh110 at thymidine residue 17 (bold and
italic) using an amino-modified C6 dT from Glenresearch with
a donor-acceptor distance of four intervening base pairs. The
complementary (+)-strand 3′-d(GCT AGC TAG GAG TAA
TAT TCA)-Cy5 was labeled with Cy5 via a C6 aminolinker
at its 5′ end. An oligonucleotide of the same sequence without
Cy5 was used as a complementary strand for control measurements of FRET-inactive conjugate. The aqueous measurement
buffer of the FRET conjugates (≈10-12 M) contained 180 mM
NaCl, 10 mM NaH2PO4/Na2HPO4, and 400 µM ascorbic acid.
The cross-correlation analysis of Figure 7B was performed
with a 62-base-pair DNA strand (Evotec, Hamburg, Germany),
which was labeled with rhodamine green (RhGr, excitation
maximum 500 nm, emission maximum 525 nm; Molecular
Probes) and MR121.
Fluorescence Detection. All experiments were performed
with a confocal epi-illuminated fluorescence microscope (IX70
microscope; Olympus, Tokyo, Japan), which has been described
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2985
previously.24,29,30 An argon ion laser (Innova Sabre; Coherent,
Palo Alto, CA) tuned to 502 or 496 nm was used for excitation
of donor fluorescence and FRET either in the continuous wave
(cw) or actively mode-locked pulsed mode (APE, Berlin,
Germany; pulse length ≈180 ps, repetition rate 73.5 MHz).
Either a cw krypton ion laser (Innova 400; Coherent) tuned to
647 nm or a pulsed diode laser at 633 nm (PDL 800; Picoquant,
Berlin, Germany; pulse length ≈100 ps, repetition rate synchronized to the actively mode-locked argon ion laser) was
employed in the case of direct excitation of fluorescence from
the red-emitting or acceptor dye. The laser light was coupled
into the microscope by a dichroic mirror (502/636PC; AHF
Analysentechnik, Tuebingen, Germany). A water immersion
microscope objective (Olympus UplanApo, 60×, N.A. 1.2) was
used to focus the laser light into the sample solution and to
collect the fluorescence light. The fluorescence light was imaged
onto the confocal pinhole and divided into parallel and
perpendicular polarized beams and then into two different
wavelength ranges below and above 620 nm (dichroic mirror
620DCXR; AHF), and each component was detected by a
separate avalanche photodiode (APD-AQR 15; Perkin-Elmer
Optoelectronics, Fremont, CA) with separate fluorescence filters
(HQ535/50 and HQ730/140; AHF). For each wavelength range
(below 620 nm (HQ535/50) denoted green and above 620 nm
(HQ730/140) denoted red), the fluorescence photons detected
for parallel and perpendicular polarization were combined. Thus,
fluorescence is monitored in two detection channels, the green
detection channel for the fluorescence of the donor dye (denoted
by the index “g”) and the red detection channel for the
fluorescence of the acceptor dye (denoted by the index “r”).
The single-photon detection events were stored by a PC card
(SPC 432; Becker & Hickl GmbH Berlin, Germany), which
recorded the arrival time since the last photon, the detection
channel (“g” or “r”), and in the case of pulsed excitation the
microscopic time relative to the next laser pulse for timecorrelated single-photon counting (TCSPC) analysis. For singlemolecule studies, photons from fluorescence bursts were
distinguished from the background signal of 1-2 kHz by
applying certain threshold criteria, which have been described
in detail before.21,24,28,67 Correlation data were either software
calculated from the photon stream of all selected single-molecule
fluorescence bursts or set up by a hardware correlator PC card
(5000E, ALV, Langen, Germany) for FCS bulk studies.
All experiments were carried out at room temperature on
freely diffusing molecules in an open detection volume. The
focal irradiance, I0, was calculated from the average excitation
power, P, and the focal 1/e2 radius, ω0, as determined by FCS
of an aqueous Rh110 solution (D ) 3 × 10-6 cm2/s,68 see eq
12) at low excitation irradiance; I0 ) P/(π/2ω02). The transit
time, tt, of a molecule was calculated from its characteristic
diffusion time, τDiff, determined by FCS; tt ) 4/3τDiff.6,15,48,49
Results and Discussion
Photobleaching of Red-Emitting Dyes by Green Excitation
Light. First, we analyzed the influence of simultaneous red and
green excitation light on the fluorescence properties of several
red-emitting dyes in aqueous solution. Laser wavelengths of
496 or 502 and 647 nm were chosen as green and red excitation
light, respectively. We investigated the free dyes Cy5, MR121,
Alexa660, and Alexa680 as well as the dyes Atto647N and
Atto655 bound to DNA (denoted Atto647N-DNA and Atto655DNA). It should be noted that fluorescence is solely elicited by
the red light. Direct excitation by green light is negligible. Figure
2A shows the abrupt decrease of the fluorescence count rate of
2986 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Eggeling et al.
Figure 2. Two-color photobleaching observed by FCS. (A and B) Fluorescence signal of an aqueous Cy5 solution (∼10-9 M) excited with 647
nm cw laser light (I0 ) 110 kW/cm2) under the addition of 502 nm cw laser light: (A) intermittent addition of 502 nm light (I0 ) 180 kW/cm2);
(B) continuous addition of 502 nm with increasing irradiance, I0 ) 0 kW/cm2 (black line), I0 ) 180 kW/cm2 (gray line), and I0 ) 360 kW/cm2
(dots). (C) FCS data resulting from the autocorrelation of the count-rate traces of part B (I0 ) 0 kW/cm2 (crosses), I0 ) 180 kW/cm2 (open dots),
and I0 ) 360 kW/cm2 (black squares)). The autocorrelation curves are normalized to the curve at 0 kW/cm2 irradiance. A fit of eq 13 to these data
regarding two dark states (gray lines) results in the following parameters. I0 ) 0 kW/cm2: c ) 1.8, τDiff ) 0.45 ms, ωz/ω0 ) 3, FD ) 0.24, and τDark
) 1.0 µs for the first dark state and FD ) 0.50 and τDark ) 0.26 µs for the second dark state, A ) 0. I0 ) 180 kW/cm2: c ) 1.7, τDiff ) 0.45 ms
(fixed), ωz/ω0 ) 3 (fixed), FD ) 0.25, and τDark ) 1.2 µs for the first dark state and FD ) 0.53 and τDark ) 0.23 µs for the second dark state, A )
0.55, kz ) 2440 s-1. I0 ) 360 kW/cm2: c ) 2.1, τDiff ) 0.45 ms (fixed), ωz/ω0 ) 3 (fixed), FD ) 0.24, and τDark ) 1.3 µs for the first dark state
and FD ) 0.51 and τDark ) 0.30 µs for the second dark state, A ) 0.65, kz ) 5000 s-1. (The experimental conditions were the following: ω0 ) 0.7
µm (τDiff(Rh110) ) 0.4 ms).) (D and E) Two-color photobleaching of MR121 (D) and Alexa660 (E). (upper panel) Fluorescence signal of an
aqueous MR121 or Alexa660 solution (∼10-9 M) excited with 647 nm cw laser light (I0 ) 170 kW/cm2 for MR121 and I0 ) 70 kW/cm2 for
Alexa660) under the intermittent addition of 496 nm cw laser light (I0 ) 250 kW/cm2). (lower panel) FCS data resulting from the autocorrelation
of the count-rate traces of the aqueous MR121 or Alexa660 solution with simultaneous illumination by 647 nm light (I0 ) 170 kW/cm2 for MR121
and I0 ) 70 kW/cm2 for Alexa660) and 496 nm light (I0 ) 0 kW/cm2 (black line), I0 ) 140 kW/cm2 (open dots), and I0 ) 250 kW/cm2 (gray line)).
The autocorrelation curves are normalized to the curve at 0 kW/cm2 irradiance.
an aqueous Cy5 solution excited by red cw light upon the
addition of green cw light. The decrease is enhanced with the
green irradiance, as depicted in Figure 2B. This characteristic
is common for all other red-emitting dyes investigated here, as
exemplarily shown in Figure 2D and E for the free dyes MR121
and Alexa660. To investigate the origin of this fluorescence
decrease further, FCS measurements were performed. Figure
2C shows autocorrelation data of Cy5 fluorescence recorded at
a constant red irradiance and increasing green irradiance. The
decrease in the decay time of the correlation data results from
an on average shortened observation time of the Cy5 dye and
scales with the green irradiance. This feature of the correlation
data is also observed for the other red-emitting dyes (e.g., see
Figure 2D and E for MR121 and Alexa660). Due to its
dependence on the green irradiance, this effect is assigned to
an enhanced photobleaching48,49,63,69,70 by the green light.
The autocorrelation data of Cy5 (Figure 2C) and of the other
red-emitting dyes shows an additional decay in the microsecond
time range. The amplitude of this decay differs between the
different dyes examined (e.g., compare parts C-E of Figure 2)
and rises with the red excitation irradiance. It is assigned to the
population of the dye’s dark states, which is for the most part
the triplet state.62 Only in the case of Cy5, an additional decay
shows up in the sub-microsecond time range. This decay is
characteristic for many cyanine dyes, including Cy5, and is
caused by photoisomerization kinetics between a dark cis state
and a fluorescent trans state.64,65
A fit of eq 13 to the FCS data yields values of the equilibrium
fraction of the dark state population, FD, the dark state
correlation time, τDark, and the effective photobleaching rate
constant, kz, and exposes details of the involved photokinetics,
as described next.
(A) Photoisomerization of Cy5 with FD ≈ 0.5 and τDark <
0.5 µs. Upon excitation, the dye Cy5 shows a fast equilibrium
between its cis state and its trans state with about 50%
population of the dark cis state. While this population remains
constant, the correlation time, τDark, decreases with increasing
red excitation irradiances. These characteristics of the photoinduced isomerization of Cy5 have in detail been studied before
using FCS.64,65 Most importantly for our experiments, the
photoisomerization-characteristic decay in the autocorrelation
data does not depend on the green irradiance. Consequently,
the green light does not seem to have an influence on the
photoisomerization kinetics of Cy5. Thus, within our photokinetic model of the dye Cy5, the photoisomerization can be
neglected and treated in terms of a reduced fluorescence
quantum yield, Φ(Cy5) ≈ 20-30%.
(B) Population of the Triplet State. The rate constants of
intersystem crossing (ISC) from S1 to T1, kISC, and of the decay
of T1 to S0, kT (see Figure 1A), can be calculated from the
Influence of Photobleaching in FRET Experiments
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2987
TABLE 1: Results of the Two-Color FCS Measurements;
Photokinetic Constants Assigned to the Red-Emitting Dyes
MR121
Cy5
Alexa660
Alexa680
Atto647N-DNA
Atto655-DNA
σ01a
τb
kISC c
kT c
σbXnd
e
σeff
bn
3.9
9.6
4.4
4.3
5.7
4.2
1.9
1.0
1.2
1.2
4.25
2.75
0.09
1.1
5.6
3.1
0.12
0.21
0.60
0.50
0.57
0.30
0.12
0.10
0.04
0.10
0.12
0.08
0.10
0.05
0.06
0.45
1.3
0.9
0.21
0.17
a
Absorption cross section at 647 nm, [10-16 cm2]. b Average
fluorescence lifetimes measured by time-correlated single-photon
counting (TCSPC), [109 s-1]; especially the decays of the linked dyes
are not single exponential. c Rate constants of ISC between S1 and T1,
kISC, and T1 and S0, kT, determined by FCS corresponding to the method
described in the text, [106 s-1]. d Composite cross section of photobleaching from Sn and Tn for green light (eq 2), [cm2 W-1 s-1]. σbXn
cannot be distinguished between the singlet (X ) S) and triplet (X )
T) systems. e Effective cross section of higher-order photobleaching for
green light (eq 6), [cm2 W-1 s-1].
determined amplitude and correlation time, FD and τDark, of the
triplet decay of the FCS data corresponding to kISC ) FD/τDark
(1 + k0/k01) and kT ) (1 - FD)/τDark.24,62 The inverse fluorescence lifetime of the dyes, k0, and the rate of excitation, k01 )
σ01I (eq 2), were determined by time-correlated single-photon
counting (TCSPC) and calculated from the applied red laser
irradiance, I, and the excitation cross section, σ01, at the applied
wavelength, respectively. Table 1 lists the resulting values of
kISC and kT as well as of k0 and σ01 of all dyes examined. The
dyes show differences of 1 order of magnitude in their rates of
ISC. The values kISC ≈ 106 s-1 and kT ≈ 105 s-1 coincide very
well with values previously determined for cyanine or rhodamine
dyes via FCS.62,64 In contrast, the dyes MR121, Atto655, and
Atto647N show almost negligible population of the triplet state
(e.g., Figure 2D), and corresponding very low values of kISC <
2 × 105 s-1. A slight increase in the triplet lifetime, kT-1, is
observed for the two dyes Atto655-DNA and Atto647-DNA,
which are bound to DNA. The triplet state of dyes dissolved in
water is to a major extent quenched by water-dissolved
oxygen.47,62 The triplet-quenching efficiency of the oxygen is
slightly decreased due to steric hindrance of the DNA.24 The
determined values of kISC and kT did not show any dependence
on the applied red or green irradiance. Thus, an influence of
photoinduced (reverse) ISC can be neglected, at least at the
irradiances applied in these measurements (<100 kW/cm2 for
red light and <400 kW/cm2 for green light). Photoinduced ISC
as well as reverse photoinduced ISC specifies the possibility
of singlet-triplet crossing via higher excited electronic
states.61,65,71-74 In contrast to our FCS measurements on the free
dye Cy5 in solution, photoinduced reverse ISC has been reported
for Cy5 attached to a peptide in poly(vinyl alcohol) (PVA)61 or
immobilized on an air-glass interface.74 It seems that the
attachment to the peptide or the PVA or air-glass environment
facilitates the photoinduced reverse ISC properties of Cy5.
(C) Photobleaching by Green Light. The values of kz from
the analysis of the FCS data increase with the green excitation
irradiance, which follows from the decrease in the decay time
of the correlation data. The rate of bleaching should, however,
in principle not depend on the green excitation irradiance, since
the green light does not directly excite the investigated redemitting dyes. The addition of green light must open up new
bleaching channels, most probably due to excitation from the
first to higher excited singlet and triplet states. These states
couple quite efficiently with ionic states in polar solvents, such
as water, and thus show a large bleaching reactivity.47-51 These
bleaching channels of higher order in irradiance have several
times been identified as the major photobleaching pathways in
confocal fluorescence microscopy and are already present at
cw excitation.47-49,63,70 This bleaching pathway is quantified by
the composite cross section for photobleaching from higher
excited states, σbXn ) σbSn or σbTn (eq 2). σbXn cannot be
distinguished between the singlet (X ) S) and triplet (X ) T)
systems, and its value is thus treated to have the same
magnitude, σbSn ) σbTn ) σbXn. Under the experimental
conditions used here, with irradiances between 100 and 500
kW/cm2, direct photobleaching from S1 and T1 is much lower
than the higher-order photobleaching. As an example, with a
typical value of σbXn ) 0.1 cm2 W-1 s-1 (see Table 1), the rate
for higher-order photobleaching at an excitation irradiance of
I ≈ 100 kW/cm2 amounts to kbXn ) σbXnI ≈ 10 000 s-1, which
is 1-2 orders of magnitudes larger than values of the rate
-1
constant, keff
b1 ) 100-1000 s , known for photobleaching
from S1 and T1.47 keff
is
thus
neglected,
and the values of σbXn
b1
can be directly calculated by eq 6 using the previously
determined photophysical parameters listed in Table 1 and the
values of kz determined from the FCS data.
All investigated dyes show a significant amount of higherorder photobleaching upon the addition of green light. The cross
sections, σbXn, lie in a range of 0.03-0.1 cm2 W-1 s-1, which
is 2 orders of magnitude larger than values of σbXn ≈ 0.0010.005 cm2 W-1 s-1, previously obtained for direct two-step
photolysis of free rhodamine 6G at 502-532 nm laser excitation.48,49 These enhanced values can be caused by several
effects: (1) Generally, red-emitting dyes have lower ionization
potentials, leading to a higher bleaching reactivity from the
higher electronic states. (2) The absorption cross section to
higher excited electronic states might be more pronounced at
around 500 nm. (3) The higher energy of a green photon
compared to a red one leads to a higher level of excitation of
Sn and Tn, closer to the ionization level and, thereby, to an
increased bleaching reactivity. Note, that we have not observed
any higher-order photobleaching due to the red excitation light,
at least not at the applied irradiances <100 kW/cm2.
Due to its relatively long lifetime (microseconds compared
to nanoseconds for the singlet state), the triplet state has in a
lot of cases been identified as the major pathway of (higherorder) photobleaching.47-49 A real comparison of the extent of
higher-order photobleaching between the different dyes therefore
has to include the triplet population, as done by the effective
cross section, σeff
bn (eq 6). The previously determined composite
cross section for photobleaching from higher excited states, σbXn
) σbSn or σbTn (eq 2), solely quantifies the efficiency of higherorder photobleaching starting from S1 or T1. In contrast, σeff
bn
represents the overall efficiency of higher-order photobleaching
after the incident of excitation to S1. The values of σeff
bn can be
calculated from σbXn applying the determined triplet parameters,
kISC and kT (eq 6 and Table 1). While the values of σbXn are
about in the same range for the different dyes, the values of the
effective cross section, σeff
bn , differ by an order of magnitude
(Table 1). σeff
bn , and thus photobleaching in general, is extremely low for those dyes that show a very low rate of
intersystem crossing, that is, MR121, Atto647N, and Atto655.
As an example, the low photobleaching reactivity of MR121
becomes obvious from the almost diminishing decrease in count
rate or FCS decay time upon the addition of green light (see,
e.g., Figure 2D). This characteristic emphasizes the important
role of the triplet state in the photostability of dyes.
The observed results show the tremendous impact of green
laser light on the photoreactivity of the red-emitting dyes. The
amount of green light should therefore carefully be chosen in
2988 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Eggeling et al.
Figure 3. Single-molecule FRET analysis. Fluorescence count rate, gF(t) and rF(t) (time resolution 1 ms), recorded in the donor detection channel
(upper panel) and in the acceptor detection channel (lower panel) for single FRET conjugates labeled with Alexa488 and Cy5 (A) and solely
labeled with Alexa488 (B). The insets of parts A and B show exemplary time trajectories of the fluorescence brightness, F1(t), determined from
consecutive bins of 100 photons within a single fluorescence burst for the donor and acceptor detection channels. (C and D) Two-dimensional
parameter histograms of fluorescence brightness pairs simultaneously detected for the donor, gF1, and acceptor, rF1, together with one-dimensional
projections (freq. ) frequency). The histograms were set up from the time trajectories, F1(t), of several single-molecule fluorescence bursts for the
FRET conjugate labeled with Alexa488 and Cy5 (5392 single-molecule events) (C) and solely labeled with Alexa488 (2656 single-molecule events)
(D). The gray lines in parts C and D mark the selection criteria for distinguishing between fluorescence bursts from the FRET-active (upper left
quadrant) and FRET-inactive (lower right quadrant) subpopulations. The area marked by the dashed gray line in part C reveals intermediate brightness values. (The experimental conditions were the following: pulsed excitation at 502 nm, I0 ) 95 kW/cm2, ω0 ) 0.6 µm (τDiff(Rh110) ) 0.3
ms).)
two-color fluorescence experiments combining simultaneous
green and red laser excitation, and in particular at or close to
saturation. The choice of irradiance is of particular importance
in FRET experiments, where the excited acceptor dye is
intrinsically exposed to green light, as characterized in the next
chapter.
Acceptor Photobleaching in FRET Experiments
Single-Molecule Analysis. Single-molecule experiments of
the FRET conjugate were performed at different irradiances of
502 nm in order to characterize the impact of the green laser
light on the acceptor photoreactivity in FRET experiments.
Figure 3 shows a typical single-molecule experiment of FRETactive (Figure 3A) and FRET-inactive (Figure 3B) conjugates
(in this case applying Alexa488 and Cy5 labels), clearly
presenting fluorescence bursts above the background signal level
due to single-molecule transits. By selecting only photons within
a fluorescence burst, that is, from a single-molecule transit, for
subsequent analysis, time trajectories of the donor and acceptor
brightness, F1(t), were set up for each single-molecule event
from a so-called sliding-photon window analysis.21,24,28,29,67
Here, the photon stream detected within a single-molecule event
was split up into subsequent windows of 100 photons each. For
each photon window, the brightness was calculated from the
corresponding length of time (F1 ) 100/length of time) and
the point in time, t, taken from the 50th photon event. The insets
to Figure 3A and B show examples of such time trajectories,
F1(t). In particular, the inset of the lower panel of Figure 3A
depicts an event of acceptor photobleaching, characterized by
an abrupt loss in acceptor brightness and an accompanying
increase in donor brightness due to the breakdown of FRET
(see Figure 1B).
It is surprising that the height of the fluorescence bursts in
the donor detection channel is the same for donor- and acceptorlabeled conjugates and for solely donor-labeled conjugates
(Figure 3A and B). The presence of fluorescence bursts in the
acceptor channel reveals the FRET activity in the case of doubly
labeled conjugates, which should lead to a decrease in the
fluorescence brightness of the donor. For a better understanding,
two-dimensional joint histograms of coincident donor and
acceptor brightness were set up from the single-molecule
trajectories, F1(t) (Figure 3C and D). Such an analysis method
has been introduced previously and enables the exposure of
species of one sample with different FRET activities.24,27,29-31
Two species are exposed in this case, that is, a FRET-active
species with high brightness, rF1, in the acceptor detection
channel and low brightness, gF1, in the donor detection channel
and a FRET-inactive species with hardly any brightness in the
acceptor detection channel and high brightness, gF1, in the donor
detection channel. Thus, FRET activity is accompanied by high
fluorescence bursts in the acceptor detection channel and
simultaneously low bursts in the donor detection channel. For
the case of FRET inactivity, this is vice versa. Intermediate cases
(dashed gray line in Figure 3C) are caused by fast acceptor
bleaching or by very rare events of multiple molecules
simultaneously traversing the detection volume. Recent work
reports photodestruction-intermediate states of an acceptor dye
that are nonemitting but still able to quench the fluorescence
of the donor.57 Such a molecule is characterized by a quenched,
low donor and no acceptor fluorescence. However, such
Influence of Photobleaching in FRET Experiments
Figure 4. Irradiance dependence of donor and acceptor brightness.
Dependence of the average brightness, F1, of single Rh110- and Cy5labeled FRET conjugates on the irradiance, I0/2, of cw (crosses) and
pulsed (black squares) 502 nm laser light: donor, D0F1 (A), and acceptor,
A
F1 (B), fluorescence. The brightness in part A was determined from
FRET-inactive fluorescence bursts, and that in part B was determined
from FRET-active fluorescence bursts, selected as outlined in Figure
3C. All data could be well described by eq 11 utilizing eq 4 in the
case of pulsed excitation and eq 7 in the case of cw excitation as well
as the corresponding photokinetic constants listed in Table 2 (gray lines).
The dashed gray line represents the simulation disregarding photoinduced (reverse) intersystem crossing and photobleaching. (The experimental conditions were the following: τDiff ) 1 ms, ω0 ) 0.6 µm
(τDiff(Rh110) ) 0.3 ms).)
photodestruction-intermediate states have not been observed in
any of the single-molecule trajectories.
As outlined, the FRET-inactive species is to a substantial
portion also present in the case of the single-molecule studies
of the donor- and acceptor-labeled conjugate. As shown in a
previous study,43 this portion appears despite intact double
labeling due to inactive or missing acceptor dye. Insufficient
hybridization of the double DNA strand can be excluded for at
least two reasons. First, potential residual single-stranded DNA
was removed during the synthesis of the conjugates. Second,
the breakup of the FRET-active double strand into two inactive
single strands after the dilution to the picomolar concentration
should lead to a rise of the fraction of FRET-inactive conjugates
over time. No such time dependence could be observed.
The gray lines in Figure 3C and D mark the selection criteria
for distinguishing between fluorescence bursts from FRETactive molecules (upper left quadrant) and FRET-inactive
molecules (lower right quadrant). By selecting only photon
bursts from the FRET-active species, the analysis is solely
focused on the truly FRET-active molecules. This artifact-free
analysis can only be achieved by single-molecule analysis, a
reason for our choice of analysis method in this case.
Irradiance Dependence of Fluorescence Brightness. Figure
4 shows the dependence of the average brightness, F1 (number
of counts per dwell time), per single-molecule burst on the
excitation irradiance of green light at 502 nm, for the donor of
the FRET-inactive species and for the acceptor of the FRETactive species, in this case of a Rh110- and Cy5-labeled FRET
conjugate. The donor and acceptor brightness coincides with
the brightness detected in the corresponding detection channels,
D0F ) gF and AF ) rF (eq 11). Cross-talk of donor
1
1
1
1
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2989
fluorescence into the acceptor detection channel, R (eq 11),
makes up less than 2% and can be disregarded for the FRETactive species (see Figure 3C). Using cw (crosses) or pulsed
(squares) excitation, the irradiance dependence of F1 shows three
typical regimes, as was also observed in previous studies:47-49,70
(1) a linear dependence for low irradiance, (2) a saturation to a
maximum value due to triplet or dark state population and
photobleaching, and (3) a pronounced decrease for high irradiances, which can only be explained by a high degree of
photobleaching reactivity from higher excited electronic states.
This decrease is exceptionally pronounced for the acceptor
fluorescence brightness in the case of pulsed excitation. All of
the data can be well described by eq 11 (utilizing eq 4 for pulsed
excitation and eq 7 for cw excitation). This description (gray
lines) uses the photophysical constants of Table 2 and introduces
products of fluorescence detection efficiency and fluorescence
quantum yield of ΨgΦD0 ) 0.0055 for the donor and ΨrΦA )
0.0027 for the acceptor in the case of cw excitation and 0.0033
in the case of pulsed excitation. Detection efficiencies in the
range of Ψ ≈ 1-5% generate reasonable values for the quantum
yields of donor, ΦD0 ≈ 0.1-0.6, and acceptor, ΦA ≈ 0.050.30, which might be slightly lower compared to the free dye
due to dark states involved when coupled to the DNA.67
(A) Irradiance Dependence of the Donor. A slight difference
is observed between pulsed (squares) and cw (crosses) excitation
for the irradiance dependence of the fluorescence brightness,
D0F , of the donor of the FRET-inactive species (Figure 4A).
1
On one hand, the maximum achievable photon counts are
slightly lower for the case of pulsed excitation. This is caused
by pulse saturation, which has in detail been outlined in previous
publications.49,69,75 Compared to cw excitation, the increased
peak irradiance of the picosecond pulses leads to a pronounced
degree of saturation of the optical S0-S1 transition, resulting in
about a 1.5-times higher achievable brightness for cw excitation.
This finding is in accordance with the factor of about 1.4
resulting from our experiments. Most prominent, however, is
photobleaching from higher excited electronic states manifested
in a more pronounced decrease of the brightness toward very
high irradiances observed for pulsed excitation. The composite
cross section, D0σbXn (eq 2), for photobleaching from higher
excited electronic states is 1 order of magnitude higher for
pulsed excitation (0.0016 cm2 W-1 s-1) than for cw excitation
(0.0001 cm2 W-1 s-1) (Table 2). Pulse enhancement of higherorder photobleaching may in general be explained by the
amplification of multistep excitation due to the significantly
enhanced photon density condensed within the pulses.49
While the absolute values of D0σbXn lie within values of σbXn
≈ 0.001-0.005 cm2 W-1 s-1, previously reported for direct
two-step photolysis of other free rhodamine dyes in aqueous
solution at 502-532 nm laser excitation,48,49 it is surprising to
note that the same reports do not observe any difference between
cw and picosecond pulsed excitation. Coupling of the rhodamine
dye to the DNA may result in different photochemical properties
such as trapping of photoelectrons.
(B) Irradiance Dependence of the Acceptor. Several other
photophysical processes have to be regarded in the case of the
FRET-induced acceptor fluorescence. First of all, FRET to the
excited singlet and triplet states has to be included besides the
energy transfer to the acceptor’s ground state (eq 8). If one only
considers FRET to the acceptor’s ground state (eq 8i), the
pronounced excitation of the acceptor at very high irradiances
will lead to a depletion of its ground state (AS0 , 1). This ground
state depletion will result in a decrease of the extent of FRET,
which is accompanied by a rise of the donor’s lifetime (see eq
2990 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Eggeling et al.
TABLE 2: Results of the FRET Photobleaching Measurements; Photokinetic Constants Assigned to the Dyes Rh110 and Cy5 in
the FRET Conjugate
σ01, Ea
τb
kISC/kTc
kbX1d
Rh110
2.36
1.1/0.1
13
Cy5
0.92
4.0j
0.33k
1.5
1.1/0.1
350
σ(Re)ISC-ne,i
0
1.0
σbXnf,i
0.0001l
0.0016m
0.005l
0.025m
/
g,i
φ(Re)ISC-n
/ h,i
φbXn
0.005
2.5
a
Absorption cross section at 502 nm, [10-16 cm2]. Cy5 is excited via FRET; thus, the energy transfer efficiency, E, is given. b Fluorescence
lifetime measured by TCSPC, [109 s-1]. c Rate constants of ISC between S1 and T1, kISC, and T1 and S0, kT, determined by FCS corresponding to the
method described in conjunction with Table 1, [106 s-1]. d Rate constant for direct photobleaching from S1 and T1, [s-1]; taken from the literature.47
e
Composite cross section of (reverse) ISC from Sn and Tn for green light (eq 2), [cm2 W-1 s-1]. f Composite cross section of photobleaching from
Sn and Tn for green light (eq 2), [cm2 W-1 s-1]. g Quantum yield of (reverse) ISC from Sn and Tn for FRET-induced excitation (eq 10c and d).
h
Quantum yield of photobleaching from Sn and Tn for FRET-induced excitation (eq 10e), [10-5]. i These parameters cannot be distinguished between
the singlet (X ) S or ISC) and triplet (X ) T or ReISC) systems. j FRET-inactive conjugate. k FRET-active conjugate. l cw excitation. m Pulsed
excitation.
10a). However, this is not observed here; a constant donor
lifetime of τD ) 0.33 ns is measured over the whole range of
excitation irradiance for the FRET-active species. With τD0 )
4 ns in the absence of FRET to an acceptor, one can estimate
the energy transfer efficiencies to ES0 ) ES1 ) ET1 ) 0.92 (eq
10a). Additional energy transfer pathways of this type have
previously been reported for single bichromophoric dye systems60 and have in particular for Cy5 been observed to be as
effective as FRET to the singlet ground state.61
Surprisingly, the experimentally determined brightness of the
acceptor is underestimated and its power dependence cannot
be described properly if we disregard any photobleaching and
assume the energy transfer efficiencies of 0.92 and values of
σ01, τ, kISC, and kT as listed in Table 2 (dashed gray line in
Figure 4B for the case of cw excitation). Our simulation may
underestimate the fluorescence brightness values for several
reasons: (1) Significant direct excitation of Cy5 by the green
light may enhance the fluorescence emission but is negligible
(<1% of the signal, i.e., Aσ01 ) 0 in eq 10b) and can be
excluded. (2) Photobleaching will result in even lower brightness
values at the saturation level. (3) Photoinduced reverse ISC is
a very likely mechanism that leads to an increase in fluorescence
at higher excitation irradiances, since it recovers dye molecules
which have been trapped in the long-living triplet state.71 This
pathway is known for cyanine dyes65 and particularly for
Cy5.61,74 Thus, to fit the data, (reverse) photoinduced ISC must
be included with effective cross sections, AσISC-n ) AσReISC-n )
1 cm2 W-1 s-1 (eq 2), for the green light and quantum yields,
Aφ /
A /
ISC-n ) φReISC-n ) 0.005 (eq 10c and d), for FRET to the
excited singlet and triplet states (see Table 2). This result is
contrary to our FCS experiments (Figure 2), where no (reverse)
photoinduced ISC has been observed for free Cy5 in solution.
As for previous studies,61,74 it seems that the attachment of the
dye to the DNA or biomolecules in general might introduce
this pronounced ISC via the higher excited electronic states.
We cannot distinguish between the singlet and triplet systems
from our data and thus cannot determine the exact real values
of the corresponding photokinetic parameters. Nevertheless, it
is useful to apply the same set of parameter values for the FRET
efficiencies to S1 and T1 and for photobleaching and photoinduced (reverse) intersystem crossing via Sn and Tn, to prove
the existence of such processes. Moreover, the obtained values
/
of AσReISC-n ) 1 cm2 W-1 s-1 and AφReISC-n
) 0.005 coincide
very well with values of photoinduced reverse intersystem
crossing previously reported for another cyanine dye.65
For high irradiances, a significant difference in acceptor
brightness is observed between cw (crosses) and pulsed (squares)
excitation (Figure 4B). Pulse saturation cannot explain
the observed decrease in brightness. Thus, photolysis from
higher electronic states needs to be considered. Different values
of cross sections, AσbXn (eq 2), and corresponding quantum
yields, Aφ/bXn (eq 10e), for photobleaching via Sn and Tn, by
excitation at 502 nm and by FRET had to be employed for a
correct description of the irradiance dependence of the bright/
ness; AσbXn ≈ 0.005 cm2 W-1 s-1 and AφbXn
) 2.5 × 10-5 for
A
the acceptor with green cw excitation, and σbXn ≈ 0.025 cm2
W-1 s-1 and Aφ/bXn ) 2.5 × 10-5 for the acceptor with green
pulsed excitation. As in the case of the donor dye, the effective
higher-order photobleaching cross section, AσbXn, is 5 times
larger in the case of pulsed excitation. In both cases, cw and
pulsed excitation, photobleaching of the donor has been included
by the corresponding values of DσbXn ) D0σbXn determined for
the donor in the absence of FRET. However, its influence is
negligible.
Cross-correlation analysis of the single-molecule time trajectories of the FRET-active conjugate was performed to further
prove the high degree of photobleaching from higher excited
states, evoked by the green excitation light.
Cross-Correlation Analysis of Single-Molecule Time Trajectories. Figure 5A and B presents the cross-correlation curves,
Ggr(tc) and Grg(tc) (eq 14), obtained from the analysis of several
single-molecule trajectories of FRET-active species, which have
been selected from bursts of the experiments with pulsed
excitation. For high irradiances, the data can be well described
by eq 15, regarding the population of dark states and acceptor
photobleaching. Since the exact theory for photobleaching
analysis of FCS experiments with pulsed excitation is very complex, it is for this work sufficient to describe the cross-correlation
data by the steady state theory assuming quasi-continuous
excitation (eqs 6 and 7). This approximation is possible because
the pulse saturation of the acceptor fluorescence is negligible,
as mentioned in the case of the data of Figure 4B.49
The cross-correlation curves show the behavior predicted from
theory (eq 15), an exponential decay in the case of Ggr(tc) and
an exponential growth in the case of Grg(tc). A fit of eq 15 to
the cross-correlation data results in values of the effective
acceptor photobleaching rate constant, Akz. As shown in Figure
5C (black squares and crosses), Akz increases with irradiance,
which is manifested in a shortened decay or rise time of the
cross-correlation data. At high irradiances, the data of Akz (black
squares) can be described by eq 6 applying values listed in Table
2 (gray line). However, this analysis is biased for irradiances
of I0 < 160 kW/cm2 (crosses), as can be observed by the change
in the decay characteristics of Grg(tc) from an exponential growth
to an exponential decay (Figure 5B). For these irradiances, the
cross-correlation data are substantially hampered by diffusion,
since the average time before bleaching, tz ) 1/Akz, is no longer
much shorter than the average transit time, tt, through the
Influence of Photobleaching in FRET Experiments
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2991
Figure 5. Acceptor photobleaching revealed by single-molecule cross-correlation analysis. Single-molecule analysis of acceptor photobleaching
for the FRET-active subpopulation of Rh110- and Cy5-labeled FRET conjugates in the case of pulsed 502 nm excitation. (A and B) Crosscorrelation curves, Ggr(tc) and Grg(tc), obtained from the analysis of fluorescence brightness time trajectories, F1(t), of several single-molecule
fluorescence bursts (between 700 and 2400 depending on the applied irradiance) for different irradiances, I0/2. (C) Values of the effective photobleaching
rate constant, Akz, of the acceptor obtained from a fit of eq 15 to the cross-correlation data of parts A and B (black squares for I0/2 > 160 kW/cm2
and crosses for I0/2 < 160 kW/cm2) and from the determination of pintact (eq 6) of part D (open squares). (D) Ratio GFRET ) rF1/gF1 of the fluorescence
brightness simultaneously detected in the acceptor and donor detection channels calculated for all single-molecule events of the FRET conjugate
for different irradiances, I0/2. The lines at GFRET ) 0.15 and GFRET ) 3 mark the threshold criteria for the distinction of FRET-inactive species
(GFRET < 0.15) and FRET-active species (GFRET > 0.15), as well as FRET-active species with and without the incident of acceptor photobleaching
(0.15 < GFRET < 3 “bleached” and GFRET > 3 “intact”). The percentages give the respective fraction of molecules. The probability pintact (eq 6)
equals the fraction of intact FRET-active molecules relative to all FRET-active molecules (42% in the upper panel and 35% in the middle panel).
Values of pintact cannot reliably be determined at high irradiances (lower panel). The thresholds in GFRET were set in a way to match values in Akz
determined by the cross-correlation analysis. (The experimental conditions were the same as those in Figure 4.)
detection volume. Thus, a correct analysis of these curves has
to take acceptor photobleaching linked to diffusion into account,
but this did not yield any reliable results. Therefore, another
analysis method was applied for low irradiances (open squares).
Counting Specific Single-Molecule Bursts: GFRET Analysis. At low irradiances, the acceptor does not bleach during each
single-molecule transit. In this case, it is feasible to count singlemolecule bursts which are characterized by a sufficiently high
ratio of GFRET ) rF1/gF1 of the fluorescence brightness
simultaneously detected in the acceptor and donor detection
channels. Analyzing all single-molecule events of the FRET
conjugate for different irradiances, Figure 5D depicts the
frequency of the bursts for certain values of GFRET. As in Figure
3C, two species are detectable: a FRET-inactive species for
GFRET < 0.15 and a FRET-active species for GFRET > 0.15. By
setting a threshold at GFRET ) 3, we can distinguish between
FRET-active molecules just undergoing acceptor photobleaching
during the transit (0.15 < GFRET < 3) and FRET-active
molecules without acceptor photobleaching (GFRET > 3). High
values of GFRET result from an on average high value of acceptor
brightness, indicating no or at least very little acceptor photobleaching.
The fraction of stable molecules (GFRET > 3) establishes the
survival probability, pintact (eq 6), of still being fluorescent after
the transit time, tt. A reliable determination of pintact, however,
is only possible for I0/2 < 160 kW/cm2, permitting the
complementary determination of Akz to the cross-correlation
analysis (open squares in Figure 5C). At higher irradiances,
bleaching occurs too early during the transit (lower panel of
Figure 5D), resulting in a very low number of intact molecules
(GFRET > 3). In this case, pintact cannot be calculated accurately.
The theoretical description of the irradiance dependence of
(eq 6) in Figure 5C obtained from the complementary crosscorrelation (black squares) and GFRET analysis (open squares)
employs the same photophysical model and constants as before
(Table 2) and has to include photobleaching from higher excited
singlet and triplet states of the acceptor with a cross section of
Aσ
2
-1 s-1. This value coincides well with
bXn ) 0.033 cm W
2
the value 0.025 cm W-1 s-1 following the analysis of the
irradiance dependence of the acceptor’s fluorescence brightness
of Figure 4D.
Conclusion of the Single-Molecule Acceptor Photobleaching Experiments. The consistent results obtained from the
different analysis methods demonstrate that both the maximum
accessible fluorescence emission and the observation time in
FRET experiments are restricted by photobleaching from higher
excited states. The cross sections, AσbXn (eq 2), for photobleaching from these electronic states of the acceptor are more than 1
order of magnitude larger than the corresponding cross sections
of the donor or of rhodamine dyes in general48,49 and correspond
very well with the values of σbXn deduced from the two-color
FCS measurements (see Table 1). Assuming an absorption cross
section of AσT1n ≈ 10-16 cm2 (eq 1) for, for example, the triplet
state, as previously reported for Cy5,61 the quantum yield of
photobleaching yields AφbXn ≈ 2.5 × 10-5 for cw excitation
and AφbXn ≈ 1.3 × 10-4 for pulsed excitation (see eq 2). While
the value of this quantum yield for cw excitation has about the
same magnitude as that following FRET, Aφ/bXn, or that of
rhodamine dyes,48,49 the photobleaching reactivity is enhanced
by a factor of 5 in the case of picosecond pulsed excitation. As
for the mere photobleaching of the donor, this pulse enhanceAk
z
2992 J. Phys. Chem. A, Vol. 110, No. 9, 2006
Figure 6. Tg - Tr analysis of the Alexa488- and Cy5-labeled FRET
conjugate excited by pulsed laser light at 496 nm with an irradiance of
I0 ) 55 kW/cm2 (A) and I0 ) 110 kW/cm2 (B); joint two-dimensional
parameter histogram and the respective one-dimensional projections
of the time difference Tg - Tr and the ratio GFRET ) rF1/gF1 determined
from several single-molecule bursts (1559 in part A and 5630 in part
B). Tg and Tr represent the average values of the macroscopic times of
all photons detected within a fluorescence burst for the donor and
acceptor detection channels, respectively. The macroscopic time depicts
the time of the detection of a photon with respect to the beginning of
the fluorescence burst (see Figure 1B). The gray box selects singlemolecule events with pairs of values (Tg - Tr, GFRET) exhibiting acceptor
photobleaching (153, i.e., 10%, in part A and 957, i.e., 17%, in part
B). (The experimental conditions were the following: ω0 ) 0.55 µm
(τDiff(Rh110) ) 0.25 ms).)
ment of higher-order photobleaching may in general be explained by the amplification of multistep excitation due to the
significantly enhanced photon density condensed within the
pulses. It is surprising that picosecond-long pulses already cause
such a gain in photobleaching, since previous studies revealed
such pronounced pulse effects only for femtosecond-long
pulses.49 In addition to a slight decline in the maximal achievable
fluorescence brightness due to pulse saturation, this strong pulse
enhancement of photobleaching hampers the use of (even
picosecond) pulsed excitation light to FRET experiments.
Exposure of Significant Acceptor Photobleaching in
Single-Molecule Experiments: Tg - Tr Analysis
The Tg - Tr analysis is an easily conceivable method for
analyzing the amount of acceptor photobleaching in singlemolecule FRET or two-color experiments in general. The present
setup permits the tagging of each fluorescence photon event of
a single-molecule transit with its macroscopic time relative to
the beginning of the transit, as exploited to build up the singlemolecule time trajectories of Figure 3. The average value of
the macroscopic time tags of all photons detected within a
fluorescence burst for the donor and acceptor detection channels,
Tg and Tr, respectively, gives the center of the observation time
of donor and acceptor fluorescence (Figure 1B).
Figure 6 depicts two-dimensional joint histograms of pairs
of values of the ratio GFRET ) rF1/gF1 of the fluorescence
brightness simultaneously detected in the acceptor and donor
Eggeling et al.
detection channels and Tg - Tr obtained from single-molecule
transits of the FRET conjugate (in this case Alexa488- and Cy5labeled) for two different irradiances. Four features are visible.
(1) FRET-active and -inactive species are distinguishable by
their value of GFRET (as has been outlined in Figure 5D). FRET
activity results in high fluorescence brightness, rF1 ) AF1, of
the acceptor and thus high values of GFRET, while FRET
inactivity shows no acceptor emission, rF1 ≈ 0, leading to low
values of GFRET. Without the incident of acceptor photobleaching, both, acceptor and donor, give rise to a homogeneous
fluorescence emission in time throughout the transit and thus
to about the same values in Tg and Tr, that is, Tg - Tr ≈ 0.
Even for the case of FRET inactivity, the signal in the acceptor
channel due to cross-talk or background signal should be
distributed homogeneously in time within the burst.
(2) Very rare events of the transit of several molecules within
one fluorescence burst (accidental multimolecule events) expose
intermediate values of GFRET and larger values of Tg - Tr that
are symmetrically distributed around zero, because multimolecule events might be due to the simultaneous transit of, for
example, FRET-active and -inactive species.
(3) Single-molecule events expressing intermediate values of
GFRET and positive, asymmetric distributed values of Tg - Tr
> 0 are characteristic for acceptor photobleaching (gray box in
Figure 6). Intermediate values of GFRET have already been
utilized for the calculation of pintact in Figure 5D. On the other
hand, the photobleaching of the acceptor dye shortens the time
period of its fluorescence emission and leads to a simultaneous
rise in the fluorescence photon emission of the donor. As
sketched in Figure 1B, such an event leads to higher values in
Tg and shortened times Tr. Thus, a positive value of the
difference Tg - Tr > 0 readily expresses the event of acceptor
photobleaching. The amount of single-molecule events within
this (GFRET, Tg - Tr) sector (gray lines) scales with the applied
irradiance and quantifies the extent of acceptor photobleaching
(10% in Figure 6A and 17% in Figure 6B).
(4) Excessive photobleaching of the acceptor furthermore
results in a lower brightness of the acceptor, rF1 ) AF1 (see
Figure 4), and thus, GFRET ) rF1/gF1, as can be observed from
the comparison of parts A and B of Figure 6.
In this case, we have constituted a slightly modified FRET
conjugate with the dye Alexa488 as the donor and pulsed
excitation at 496 nm to exemplify that extensive acceptor
photobleaching is not a problem specific to the previously
applied FRET conjugate at 502 nm laser excitation.
The Tg - Tr analysis thus enables the optimization of
experimental conditions with respect to photobleaching. The
increased amounts of events in the marked (GFRET, Tg - Tr)
sector (gray box in Figure 6) reveal a significant increment in
acceptor photobleaching at higher irradiances. In addition to
the adjustment of the irradiance, the exposed excess of photobleaching can be minimized by the addition of certain chemicals
such as ascorbic acid. The favorable stabilization effect of
ascorbic acid is well-known, especially for the case of photolysis
from higher excited electronic states involving ionic states,47,70,76
a reason for the addition of ascorbic acid to the buffered solution
of FRET conjugate from the initial stages of our experiments.
We have to note that the Tg - Tr analysis is not confined to
FRET experiments but enables one to readily check the amount
of photobleaching of any of the involved dyes in multicolor
experiments. More generally speaking, this method can directly
characterize the coincidence of the fluorescence signal in
multicolor experiments and thus directly expose the existence
of multiple species or conformational states of a sample.
Influence of Photobleaching in FRET Experiments
Figure 7. Optimization of pulsed multicolor experiments. (A and B)
Fluorescence signal of an aqueous Cy5 solution (∼10-9 M) excited
with 633 nm pulsed laser light (I0 ) 90 kW/cm2) under the intermittent
addition of 496 nm pulsed laser light (I0 ) 160 kW/cm2) where marked
with a gray line (left panels). (right panels) Relative timing of the
synchronized pulses of the two lasers, 496 nm (gray line) and 633 nm
(black line); overlapping pulse timing (A) and green-preceding-red pulse
timing (B). The pulse timing was recorded using TCSPC, for the 633
nm laser via the time-resolved fluorescence decay of Cy5 in the red
detection channel (black line) and for the 496 nm laser via scattering
light detected in the green detection channel (gray line). (C) Crosscorrelation function, Grg(tc), of the fluorescence signal from the red
and green detection channels of a 62-base-pair DNA, doubly labeled
with RhGr and MR121, for simultaneous excitation by synchronized
pulsed laser light of 496 and 502 nm with an overlapping pulse timing
(gray line) and green-preceding-red pulse timing (black line). (The
experimental conditions were the following: ω0 ) 0.6 µm (τDiff(Rh110)
) 0.3 ms).)
Moreover, Tg - Tr analysis is a very useful tool in the burst
analysis of the multiparameter fluorescence detection (MFD).
Those molecules which are identified in the Tg - Tr analysis as
events with acceptor photobleaching or multimolecule events
can be easily excluded from the subsequent selective burst
analysis.
Minimization of Photobleaching in Multicolor
Experiments Applying Pulsed Excitation
In the previous chapters, we have demonstrated the deteriorating impact of using pulsed excitation in FRET experiments
caused by acceptor photobleaching. As outlined in Figure 2 and
Table 1, the simultaneous use of green and red excitation in
general obscures the fluorescence emission of a red-emitting
dye. This feature is therefore particularly characteristic of twocolor experiments constituting simultaneous green and red
pulsed laser light. This is exemplified in Figure 7 for the case
of an aqueous Cy5 dye solution with a synchronized pulsed
excitation at 496 and 633 nm. The coinciding occurrence of
pulsed green and red laser light leads to a drop in Cy5
fluorescence emission due to the previously characterized
excessive photobleaching from higher excited electronic states
exposed by the green light (Figure 7A). However, synchronizing
the two pulses such that the green laser pulse directly precedes
the red laser pulse can minimize this photobleaching, as is
evident from the vanishing drop in fluorescence following
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2993
simultaneous illumination by green light (Figure 7B). With a
repetition rate of 73.5 MHz, the green pulse in this case follows
about 11.1 ns after the previous red laser pulse. Since the
fluorescence lifetime of Cy5 is 1.0 ns, the excited Cy5 molecules
have at the incident of the green laser pulse to the most part
relaxed back to the ground state, scaling the excitation to higher
electronic states and thus photobleaching to a minimum amount.
Once again, we have to note that the direct excitation of Cy5
by the green laser light is negligible. Thus, multicolor fluorescence experiments can be efficiently improved when applying
the appropriate timing of the applied laser pulses. However,
this optimized timing will be less effective in the case of efficient
triplet buildup. With a lifetime of microseconds, the triplet state
will hardly decay until the arrival of the next pulse (11.1 ns).
In the present case of Cy5, triplet buildup is kept rather low
due to the applied irradiance (<90 kW/cm2).
The improvement gained by the appropriate timing of the
applied laser pulses is exemplified in Figure 7C, where the
coinciding fluorescence elicited from a 62-base-pair DNA
strand, doubly labeled with RhGr and MR121, has been
investigated using simultaneous 496 and 633 nm pulsed
excitation. The amount of coinciding signal is best quantified
by the cross-correlation function (eq 13) calculated from the
simultaneously detected RhGr and MR121 fluorescence; a
higher amount of coinciding signal results in an increased
amplitude of the corresponding cross-correlation function.18
Enhanced photobleaching of one of the dyes will thus lead to
a decrease in the cross-correlation amplitude. This can be
extracted from Figure 7C, where the pulses of green and red
laser light have been synchronized to the same point in time
(gray line) and to the green pulse directly preceding the red
pulse (black line). The optimized timing of the laser pulses
minimizes the photobleaching of the red dye, maximizes the
cross-correlation amplitude, and results in minimally biased
results. As outlined above, this feature is optimized by the almost
vanishing triplet population of MR121 (kISC/kT ) 0.3, Table
2).
Conclusions
We have presented different analysis tools based on singlemolecule detection to show that in multicolor fluorescence
experiments, including FRET, the performance of the longwavelength dye is limited by photobleaching, which is even
more pronounced by the blue-shifted laser light. The different
analysis tools include fluorescence correlation spectroscopy
(FCS) of bulk solutions, selective fluorescence cross-correlation
of single-molecule trajectories, and multidimensional histogram
analysis of different fluorescence parameters detected for singlemolecule events (multiparameter fluorescence detection (MFD)).
Taking several red-emitting dyes as an example, the maximum
achievable fluorescence brightness as well as observation time
is exceptionally limited due to the photobleaching caused by
the green laser light of around 500 nm. Exerting an impact on
the already excited fluorophore, the green laser light effectively
induces the population of higher electronic singlet and triplet
states. In polar solvents such as water, these states couple quite
efficiently with ionic states and lead to a distinct bleaching
reactivity. This not only is the case for simultaneous red and
green laser illumination but also has a particular influence on
FRET experiments. In FRET experiments, the acceptor dye is
intrinsically exposed to green laser light, introducing enhanced
photobleaching, per se. The enhanced photobleaching from
higher excited electronic states is particularly distinctive for the
case of pulsed excitation. Due to the presence of multiple
2994 J. Phys. Chem. A, Vol. 110, No. 9, 2006
absorption steps, the high density of excitation photons present
in the short laser pulses intensifies the excitation to the higher
excited electronic states. Aside from a slight decline in the
maximum achievable fluorescence brightness due to pulse
saturation, this strong pulse enhancement of photobleaching
limits the use of pulsed excitation light to FRET experiments
already for the case of picosecond-long pulses. As a consequence, the performance of multicolor fluorescence experiments
applying simultaneously green and red pulsed laser light can
be efficiently improved when using the correct timing for the
laser pulses, that is, the green pulse directly preceding the red
pulse. A readily applicable method to expose the extent of such
photobleaching has been introduced by calculating the difference
of average photon times of donor or green fluorophore
fluorescence emission and acceptor or red fluorophore fluorescence emission for a single-molecule fluorescence burst. This
Tg - Tr analysis not only enables one to readily check and
optimize the performance of multicolor and FRET experiments
with regard to photobleaching, but it also in general manages
to unmask the presence of different species. Moreover, Tg - Tr
analysis is an important tool to exclude those molecules from
the subsequent selective burst analysis, which have been
identified as events with acceptor photobleaching or multimolecule events.
Nevertheless, higher-order photobleaching and pulse saturation are only present at rather high excitation irradiances, as is
the case in confocal microscopy. High irradiances are necessary
to produce a significant population of the dye’s excited states,
which enables the additional absorption steps to the higher
excited electronic states. Photolysis in experiments applying
lower levels of light, such as wide-field microscopy, is solely
given by direct photobleaching from the first excited singlet
state and from the lowest triplet state, resulting in higher
photostabilities and, thus, longer observation times but lower
fluorescence brightness.
Potential improvements introduced by, for example, moving
to other blue or green wavelengths in multicolor or FRET
experiments would most likely not introduce much improvement. The absorption spectra of the first excited singlet state or
the lowest triplet state are usually rather broad, revealing no
significant difference in the absorption cross section between
the potential selectable wavelengths. Furthermore, more blueshifted laser light would lead to excitation to even higher excited
electronic states with more pronounced photobleaching reactivity. In contrast, more red-shifted laser light introduces a
significant extent of excitation cross-talk, that is, direct excitation
of the acceptor or red-emitting dye, which should be kept as
low as possible in multicolor experiments.
Previous experiments simultaneously applying green and red
excitation of an aqueous Cy5 solution observed a return of
fluorescence by green light on single Cy5 molecules previously
photobleached by the red light.58,59 These characteristics do not
contradict our results, inasmuch as the applied irradiances were
much lower than in our case (1 W/cm2 up to 10 kW/cm2
compared to >100 kW/cm2) as well as the experimental
conditions significantly different. The photoswitching behavior
was indeed only observable under conditions of efficient oxygen
removal as well as with the addition of a triplet quencher. This
might indicate the significant role of oxygen as well as of the
triplet state in the photoreactivity of the fluorophore, as depicted
in our results and in earlier studies.47,49,70 Further, we observed
significant photoinduced (reverse) intersystem crossing for the
dye Cy5 attached to the DNA. Similar photoinduced (reverse)
intersystem crossing has been reported for Cy5 attached to a
Eggeling et al.
peptide61 or immobilized on an air-glass interface.74 In contrast,
the free dye Cy5 in solution did not show such a characteristic.
These findings emphasize the sensitivity of a dye’s photophysics
and thus fluorescence emission on environmental properties.
The presented experiments prove the superior capabilities of
single-molecule experiments. Bulk measurements on the present
FRET probes would have resulted in biased results due to a
substantial portion of FRET-inactive species present. Only
single-molecule analysis offers the capabilities required to
constrain the analysis on the relevant fluorescence signal.
Nevertheless, the disclosed observations demand a careful
adjustment of the applied laser irradiances, excitation mode, and
pulse timings in multicolor and FRET experiments. Singlemolecule methods such as the cross-correlation or Tg - Tr
analysis help one to readily access optimized experimental
conditions.
Acknowledgment. This paper is dedicated to Professor
Jürgen Troe on the occasion of his 65th birthday to honor his
support and his numerous scientific contributions. We thank S.
Höppner for critically reading the manuscript and A. Valeri and
D. Pfiffi for the determination of certain dye properties. C.E.
thanks S. Hell for his support. Financial support by the BMBF
Biofuture grant 0311865, BMBF grant 0312020A, and the SFB
663, Heinrich-Heine-Universität, is gratefully acknowledged.
J.W. acknowledges support from the Swedish Research Council
and the Swedish Foundation for International Cooperation in
Research and Higher education.
Supporting Information Available: Description of the
cross-correlation analysis of single-molecule FRET acceptor
photobleaching. This material is available free of charge via
the Internet at http://pubs.acs.org.
References and Notes
(1) Lakowicz, J. R. Principles of fluorescence spectroscopy; Plenum
Press: New York, 1999.
(2) Moerner, W. E.; Kador, L. Phys. ReV. Lett. 1989, 62, 2535.
(3) Orrit, M.; Bernard, J. Phys. ReV. Lett. 1990, 65, 2716.
(4) Shera, E. B.; Seitzinger, N. K.; Davis, L. M.; Keller, R. A.; Soper,
S. A. Chem. Phys. Lett. 1990, 174, 553.
(5) Rigler, R.; Mets, Ü.; Widengren, J.; Kask, P. Eur. Biophys. J. 1993,
22, 169.
(6) Eigen, M.; Rigler, R. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 5740.
(7) Nie, S.; Chiu, D. T.; Zare, R. N. Anal. Chem. 1995, 67, 2849.
(8) Zander, C.; Sauer, M.; Drexhage, K. H.; Ko, D. S.; Schulz, A.;
Wolfrum, J.; Brand, L.; Eggeling, C.; Seidel, C. A. M. Appl. Phys. B 1996,
63, 517.
(9) Xie, X. S.; Trautmann, J. K. Annu. ReV. Phys. Chem. 1998, 49,
441.
(10) Deniz, A. A.; Dahan, M.; Grunwell, J. R.; Ha, T.; Faulhaber, A.
E.; Chemla, D. S.; Weiss, S.; Schultz, P. G. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 3670.
(11) Ambrose, W. P.; Goodwin, P. M.; Jett, J. H.; van Orden, A.; Werner,
H. J.; Keller, R. A. Chem. ReV. 1999, 99, 2929.
(12) Tinnefeld, P.; Sauer, M. Angew. Chem., Int. Ed. 2005, 44, 2642.
(13) Fluorescence Correlation SpectroscopysTheory and Applications;
Rigler, R., Elson, E. L., Eds.; Springer: Berlin, Heidelberg, 2001.
(14) Magde, D.; Elson, E. L.; Webb, W. W. Phys. ReV. Lett. 1972, 29,
705.
(15) Fries, J. R.; Brand, L.; Eggeling, C.; Köllner, M.; Seidel, C. A. M.
J. Phys. Chem. A 1998, 102, 6601.
(16) Chen, Y.; Müller, J. D.; So, P. T. C.; Gratton, E. Biophys. J. 1999,
77, 553.
(17) Kask, P.; Palo, K.; Ullmann, D.; Gall, K. Proc. Natl. Acad. Sci.
U.S.A. 1999, 96, 13756.
(18) Schwille, P.; Meyer-Almes, F. J.; Rigler, R. Biophys. J. 1997, 72,
1878.
(19) Kask, P.; Palo, K.; Fay, N.; Brand, L.; Mets, Ü.; Ullmann, D.;
Jungmann, J.; Pschorr, J.; Gall, K. Biophys. J. 2000, 78, 1703.
(20) Chen, Y.; Tekmen, M.; Hillesheim, L.; Skinner, J.; Wu, B.; Müller,
J. D. Biophys. J. 2005, 88, 2177.
Influence of Photobleaching in FRET Experiments
(21) Eggeling, C.; Berger, S.; Brand, L.; Fries, J. R.; Schaffer, J.;
Volkmer, A.; Seidel, C. A. M. J. Biotechnol. 2001, 86, 163.
(22) Kühnemuth, R.; Seidel, C. A. M. Single Mol. 2001, 2, 251.
(23) van Orden, A.; Machara, N. P.; Goodwin, P. M.; Keller, R. A.
Anal. Chem. 1998, 70, 1444.
(24) Eggeling, C. Ph.D. Thesis, Georg-August-Universität Göttingen,
1999 (in German).
(25) Prummer, M.; Hübner, C. G.; Sick, B.; Hecht, B.; Renn, A.; Wild,
U. P. Anal. Chem. 2000, 72, 443.
(26) Herten, D. P.; Tinnefeld, P.; Sauer, M. Appl. Phys. B 2000, 71,
765.
(27) Deniz, A. A.; Laurence, T. A.; Dahan, M.; Chemla, D. S.; Schultz,
P. G.; Weiss, S. Annu. ReV. Phys. Chem. 2001, 52, 233.
(28) Eggeling, C.; Schaffer, J.; Volkmer, A.; Seidel, C. A. M.; Brand,
L.; Jäger, S.; Gall, K. Proc. Biosensor 2001, http://w210.ub.uni-tuebingen.de/
portal/BioSensor2001.
(29) Margittai, M.; Widengren, J.; Schweinberger, E.; Schröder, G. F.;
Felekyan, S.; Haustein, E.; König, M.; Fasshauer, D.; Grubmüller, H.; Jahn,
R.; Seidel, C. A. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 15516.
(30) Rothwell, P. J.; Berger, S.; Kensch, O.; Felekyan, S.; Antonik, M.;
Wöhrl, B. M.; Restle, T.; Goody, R. S.; Seidel, C. A. M. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 1655.
(31) Kapanidis, A. N.; Lee, N. K.; Laurence, T. A.; Doose, S.; Margeat,
E.; Weiss, S. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 8936.
(32) Widengren, J.; Kudryavtsev, V.; Antonik, M.; Berger, S.; Gerken,
M.; Seidel, C. A. M. Anal. Chem., in press.
(33) Diez, M.; Zimmermann, B.; Börsch, M.; König, M.; Schweinberger,
E.; Steigmiller, S.; Reuter, R.; Felekyan, S.; Kudryavtsev, V.; Seidel, C.
A. M.; Gräber, P. Nat. Struct. Mol. Biol. 2004, 11, 135.
(34) Laurence, T. A.; Kapanidis, A. N.; Kong, X.; Chemla, D. S.; Weiss,
S. J. Phys. Chem. B 2004, 108, 3051.
(35) Förster, T. Ann. Phys. 1948, 2, 55.
(36) Clegg, R. M. Methods Enzymol. 1992, 211, 353.
(37) Van Der Meer, B. W.; Coker, G. I.; Chen, S. Y. S. Resonance
Energy Transfer: Theory and Data; VCH Publishers: New York,
1994.
(38) Resonance Energy Transfer; Andrews, D. L., Demidov, A. A., Eds.;
John Wiley & Sons: Chicester, U.K., 1999.
(39) Ha, T. Methods 2001, 25, 78.
(40) Jares-Erijman, E. A.; Jovin, T. M. Nat. Biotechnol. 2003, 21,
1387.
(41) Sekar, R. B.; Periasamy, A. J. Cell Biol. 2003, 160, 629.
(42) Kohl, T.; Heinze, K. G.; Kuhlemann, R.; Koltermann, A.; Schwille,
P. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 12161.
(43) Eggeling, C.; Kask, P.; Winkler, D.; Jäger, S. Biophys. J. 2005,
89, 605.
(44) Eggeling, C.; Jäger, S.; Winkler, D.; Kask, P. Curr. Pharm.
Biotechnol. 2005, 6, 351.
(45) Schuler, B.; Lipman, E. A.; Eaton, W. A. Nature 2002, 419, 743.
(46) Müller, B. K.; Zaychikov, E.; Bräuchle, C.; Lamb, D. C. Biophys.
J. 2005, 89, 3508.
J. Phys. Chem. A, Vol. 110, No. 9, 2006 2995
(47) Eggeling, C.; Widengren, J.; Rigler, R.; Seidel, C. A. M. In Applied
fluorescence in chemistry, biology and medicine; Rettig, W., Strehmel, B.,
Schrader, M., Seifert, H., Eds.; Springer: Berlin, 1999; p 193.
(48) Eggeling, C.; Widengren, J.; Rigler, R.; Seidel, C. A. M. Anal.
Chem. 1998, 70, 2651.
(49) Eggeling, C.; Volkmer, A.; Seidel, C. A. M. ChemPhysChem 2005,
6, 791.
(50) Anbar, M.; Hart, E. J. Am. Chem. Soc. 1964, 86, 5633.
(51) Khoroshilova, E. V.; Nikogosyan, D. N. J. Photochem. Photobiol.,
B 1990, 5, 413.
(52) Damjanovic, S.; Vereb, G.; Schaper, A.; Jenei, A.; Matko, J.;
Starink, J. P. P.; Fox, G. Q.; Arndt-Jovin, D. J.; Jovin, T. M. Proc. Natl.
Acad. Sci. U.S.A. 1995, 92, 1122.
(53) Mekler, V. M.; Averbakh, A. Z.; Sudarikov, A. B.; Kharitonova,
O. V. J. Photochem. Photobiol., B 1997, 40, 278.
(54) Wouters, F. S.; Bastiaens, P. I. H. Curr. Biol. 1999, 9, 1127.
(55) Neher, R.; Neher, E. J. Microsc. 2004, 213, 46.
(56) Zal, T.; Gacoigne, N. R. J. Biophys. J. 2004, 86, 3923.
(57) Ha, T.; Xu, J. Phys. ReV. Lett. 2003, 90, 223002.
(58) Bates, M.; Blosser, T. R.; Zhuang, X. Phys. ReV. Lett. 2005, 94,
108101.
(59) Heilemann, M.; Margeat, E.; Kasper, R.; Sauer, M.; Tinnefeld, P.
J. Am. Chem. Soc. 2005, 127, 3801.
(60) Hofkens, J.; Cotlet, M.; Vosch, T.; Tinnefeld, P.; Weston, K. D.;
Ego, C.; Grimsdale, A.; Müllen, K.; Beljonne, D.; Bredas, J. L.; Jordens,
S.; Schweitzer, G.; Sauer, M.; de Schryver, F. C. Proc. Natl. Acad. Sci.
U.S.A. 2003, 100, 13146.
(61) Tinnefeld, P.; Buschmann, V.; Weston, K. D.; Sauer, M. J. Phys.
Chem. A 2003, 107, 323.
(62) Widengren, J.; Mets, Ü.; Rigler, R. J. Phys. Chem. 1995, 99, 13368.
(63) Widengren, J.; Rigler, R. Bioimaging 1996, 4, 149.
(64) Widengren, J.; Schwille, P. J. Phys. Chem. A 2000, 104, 6416.
(65) Widengren, J.; Seidel, C. A. M. Phys. Chem. Chem. Phys. 2000,
2, 3435.
(66) Widengren, J.; Schweinberger, E.; Berger, S.; Seidel, C. A. M. J.
Phys. Chem. A 2001, 105, 6851.
(67) Eggeling, C.; Fries, J. R.; Brand, L.; Günther, R.; Seidel, C. A. M.
Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 1556.
(68) Hansen, R. L.; Zhu, X. R.; Harris, J. M. Anal. Chem. 1998, 70,
1281.
(69) Mertz, J. Eur. Phys. J. D 1998, 3, 53.
(70) Dittrich, P. S.; Schwille, P. Appl. Phys. B 2001, 73, 829.
(71) Reindl, S.; Penzkofer, A. Chem. Phys. 1996, 211, 431.
(72) Larkin, J. M.; Donaldson, W. R.; Foster, T. H.; Knox, R. S. Chem.
Phys. 1999, 244, 319.
(73) English, D. S.; Harbron, E. J.; Barbara, P. F. J. Phys. Chem. A
2000, 104, 9057.
(74) Huang, Z.; Ji, D.; Xia, A. Colloids Surf., A 2005, 257-258, 203.
(75) Gregor, I.; Patra, D.; Enderlein, J. ChemPhysChem 2004, 5, 1.
(76) Widengren, J.; Eggeling, C.; Seidel, C. A. M. Manuscript in
preparation.